q-bio.BM — Pith

0

cs.CV 2026-07-03

MolSight adds bond adjacency to vision tokens for molecular images

by Wenda Wang, Yihan Tong +2 more

MolSight: A Graph-Aware Vision-Language Model for Unified Chemical Image Understanding

Two modules for topology and grounding let the model outperform prior VLMs and chemistry tools on visual tasks.

abstract click to expand

Using molecular large language models (LLMs) as a unified framework for understanding molecular structures and functions is emerging as a new trend in tasks such as molecular design and drug discovery. However, these models struggle to fully capture the visual representation of molecular structures, limiting their potential. While existing molecular vision-language models (VLMs) show promise, they still face challenges in structural alignment and lack the necessary topological modeling for accurate molecular understanding. To address this, we propose MolSight, a graph-aware vision-language model framework designed to enhance the understanding of molecular images by VLMs. MolSight integrates a Molecular Topology Module to inject chemical-bond adjacency information into vision tokens, and a Molecular Grounding Module to align visual features with chemical symbolic semantics. Our experiments demonstrate that MolSight significantly outperforms existing VLMs, molecular LLMs, and specialized tools across multiple chemical visual understanding tasks, achieving a new level of molecular image reasoning.

0

physics.chem-ph 2026-07-02

NNPs reach near-chemical accuracy on 545-atom enzyme clusters with under 1,000 points

by Weiliang Luo, Heather J. Kulik

Enerzyme: A Framework for Efficient Training of Reactive Neural Network Potentials for Enzyme Catalysis with Application to Methyltransferases

Methyltransferase reaction energetics and transition states match DFT after training on system-specific reactive data.

abstract click to expand

Quantum mechanical (QM) cluster models provide an effective framework for mechanistic studies of enzymatic reactions but remain computationally demanding. Neural network potentials (NNPs) offer a promising route to reduce this cost, but enzymes present challenges beyond small molecules, including large system sizes, implicit-solvent environments, substantial polarization, and charge transfer. Here, we present an integrated software framework for efficient NNP training for mechanistic studies of enzymes, demonstrated on QM cluster models of S-adenosyl-L-methionine-dependent methyltransferases (MTases). Our Enerzyme code introduces modular electrostatics-aware NNP architectures and combines automated QM-cluster construction with reactive dataset generation. The Enerzymette subpackage automates reaction pathway exploration at both NNP and DFT levels. We show that iterative flexible scans and nudged elastic band calculations impose stricter requirements on NNPs than conventional dataset metrics. Nevertheless, NNPs trained on fewer than 1,000 system-specific datapoints reproduce reaction energetics and transition-state structures for MTase clusters containing up to 545 atoms with near-chemical accuracy. Direct supervision of atomic charges and consistent dielectric screening substantially improve simulation stability and accuracy, while multitask-learned atomic charges capture charge transfer and polarization trends and provide chemically meaningful descriptors of reactivity. Finally, transferability across chemically diverse catechol O-methyltransferase substrates indicates that NNPs learn generalizable reactivity patterns as training data expand across multiple enzymes. Together, these results establish a foundation for accelerating enzyme mechanistic studies and guide future NNP development for biomolecular reactivity.

0

q-bio.BM 2026-07-02

Algebraic descriptors raise B-factor accuracy 34.5% over GNM

by Honghao Zhang, Hongsong Feng

Commutative Algebra Learning for Protein Flexibility Analysis

Localized commutative-algebra features at multiple scales enable both improved prediction on 364 proteins and blind cross-protein models.

abstract click to expand

Protein flexibility, commonly quantified by B-factors, is closely related to protein structure and function. However, accurate B-factor prediction remains challenging due to the multiscale nature of protein structures and the complexity of atomic interactions. In this work, we propose a commutative algebra-based learning framework, termed CAL, for protein B-factor prediction. Unlike many biomolecular prediction tasks that rely primarily on global structural representations, B-factor prediction requires an accurate characterization of the local geometric environments surrounding individual atoms. To address this challenge, CAL employs commutative algebra theory to construct localized algebraic descriptors at multiple spatial scales. On a benchmark dataset of 364 proteins, CAL improves prediction accuracy by 34.5\% over the classical Gaussian network model (GNM). Extensive experiments demonstrate that CAL achieves robust and consistent performance across diverse datasets and is competitive with existing state-of-the-art methods. Furthermore, by integrating CAL with machine learning, we develop a blind prediction model capable of cross-protein B-factor prediction. Overall, CAL provides an effective, efficient, and mathematically principled framework for protein flexibility prediction and offers a powerful approach for analyzing and predicting localized structural properties in complex biomolecular systems.

0

cs.LG 2026-07-02

Active-GRPO raises average SRxSim to 0.1773 by updating references

by Xuefeng Liu, Mingxuan Cao +4 more

Active-GRPO: Adaptive Imitation and Self-Improving Reasoning for Molecular Optimization

The method switches from imitation to self-reinforcement per instance and replaces the reference with better policy candidates to exceed pri

abstract click to expand

Scientific reasoning is an increasingly important capability of large language models, yet improving the robustness and efficiency of training such reasoning remains a key open challenge. We study this problem in instruction-based molecular optimization, where answer-only supervised fine-tuning (SFT) collapses multi-step reasoning and reinforcement learning with verifiable rewards (RLVR) suffers from sparse feedback. Reference-guided Policy Optimization mitigates both by anchoring policy updates to dataset-provided references, but its effectiveness is tightly coupled to reference quality: weak or misaligned references impose a performance ceiling. To overcome this ceiling, we propose active reasoning, a paradigm in which the policy actively decides, on a per-instance basis, when to imitate a reference and when to reinforce its own discoveries, while continuously upgrading what it imitates. We instantiate this paradigm as Active Group Relative Policy Optimization (Active-GRPO), realized through two coupled mechanisms: active imitate-reinforce and active referencing. The former performs imitation learning when the reference still outperforms the policy's own candidates, and shifts to self-improvement via reinforcement learning once the policy has generated molecules that surpass the reference. The latter continuously upgrades the reference itself by replacing it with the best policy-generated candidate discovered so far, progressively raising the imitation target and ensuring that reference guidance remains informative-rather than restrictive-throughout training. Across TOMG-Bench MOLOPT, Active-GRPO improves average SRxSim from 0.0959 for GRPO and 0.1665 for RePO to 0.1773 under matched three-seed evaluation, with statistically significant gains on LogP, MR, and QED.

0

q-bio.BM 2026-07-01

Frustration patterns lift alternative conformation recovery over AF-Cluster

by Hanqun Cao, Zijun Gao +4 more

SF-Cluster: Frustration-Guided MSA Subsampling for Alternative Protein Conformation Recovery

Benchmark across 48 cases shows +15.5 point gain for allosteric proteins and transfer to other predictors.

abstract click to expand

Deep-learning structure predictors are sensitive to their multiple sequence alignment (MSA) input, making MSA subsampling a practical route to recovering alternative conformations. Existing approaches such as AF-Cluster operate in sequence space, providing limited control over which conformational basin is sampled. We introduce SF-Cluster, which subsamples MSAs using patterns of predicted local energetic frustration, a representation largely independent of sequence similarity. Across a benchmark of 48 cases spanning fold-switching, allosteric, oligomerization-coupled, and intrinsically disordered systems, and using an AF-Cluster-style dual-reference RMSD criterion, SF-Cluster improves target-state recovery of the alternative conformation over AF-Cluster across the two-state classes, with the largest improvement observed for allosteric systems (+15.5 percentage points). The selected MSAs transfer to an architecturally distinct predictor, indicating that the conformational signal resides in MSA composition. Mechanistically, matched-depth controls show that this recovery advantage is largely explained by the effective depth of the selected subsets, which frustration-pattern selection reliably reaches. At the same time, highly frustrated residues are enriched at sites supported by deep mutational scanning and NMR two-state exchange, and frustration covariation is enriched at state-switching contacts while remaining distinct from coevolutionary coupling. Together, these results identify frustration patterns as a transferable representation for conformational prediction and position MSA subsampling as a representation-guided reweighting problem.

0

q-bio.BM 2026-06-30

TCR model shows generated structures yield weaker contact maps

by Jiarui Li, Zixiang Yin +5 more

Structure-Regularized Interpretable TCR-Epitope Prediction

TCR-SRIM matches top accuracy yet finds experimental structures produce more accurate and diverse interaction patterns than AlphaFold3 or TC

abstract click to expand

T cell receptor (TCR)-epitope binding prediction is essential for understanding adaptive immunity and developing immunotherapies. Existing sequence- and structure-based models often generalize poorly to unseen epitopes and provide limited interpretability. Furthermore, the impact of generated structures on model learning remains unclear. We present TCR-SRIM, a structure-regularized interpretable-by-design model that combines protein language model embeddings with interpretable contact prototypes to capture residue-level TCR-epitope interactions. TCR-SRIM achieves state-of-the-art predictive performance and improved interpretation quality on the TCR-XAI benchmark. Using its inherent interpretability, we further evaluate the effect of generated structures on model learning. While structures predicted by AlphaFold3, TCRModel2, and tFold-TCR yield competitive performance, they lead to less accurate interaction patterns and reduced binding-site diversity than experimentally-resolved structures. Our results highlight limitations of current structure prediction models for TCR-epitope learning and demonstrate the value of interpretable-by-design models for studying generated biological structures.

0

q-bio.BM 2026-06-29

Manganese-GelMA hydrogels enable MRI-guided cancer immunotheranostics

by Motahareh Nazari, Keyvan Alavi

Manganese-Functionalized GelMA Hydrogels for MRI-Guided Immunotheranostics in Precision Oncology

Review outlines how these materials combine diagnosis, localized therapy, and immune modulation in one platform.

abstract click to expand

Precision oncology requires multifunctional platforms capable of integrating accurate tumor diagnosis, localized therapeutic delivery, immune modulation, and real-time monitoring of treatment response. Gelatin methacryloyl (GelMA) hydrogels have emerged as versatile biomaterials for biomedical engineering because of their biocompatibility, extracellular matrix-like structure, tunable mechanical properties, photocrosslinkability, and capacity to incorporate therapeutic agents, imaging probes, and functional nanomaterials. In parallel, manganese-based materials have gained increasing attention as promising alternatives to gadolinium-based magnetic resonance imaging contrast agents and as therapeutic components capable of modulating the tumor microenvironment. Manganese ions and manganese-based nanomaterials can enhance T1-weighted MRI contrast, generate reactive oxygen species, relieve tumor hypoxia, deplete glutathione, promote immunogenic cell death, and activate the cyclic GMP-AMP synthase-Stimulator of Interferon Genes pathway. The integration of manganese-based systems with GelMA hydrogels offers a promising strategy for developing localized, stimuli-responsive, and MRI-guided immunotheranostic platforms. This review summarizes the fundamental properties of GelMA hydrogels, the diagnostic and therapeutic roles of manganese-based materials, strategies for constructing manganese-functionalized GelMA systems, and their potential applications in precision oncology. Current challenges, including manganese-associated toxicity, controlled ion release, mechanical optimization, reproducibility, and clinical translation, are also discussed. Finally, future directions are proposed for the rational design of safe, scalable, and personalized manganese-functionalized GelMA platforms for cancer diagnosis and therapy.

0

q-bio.BM 2026-06-29

Transformers with active learning beat double-data baselines on PRRS epitopes

by Aspen Erlandsson Brisebois, Zahed Khatooni +6 more

Transformer-Based Active Learning for Data-Efficient Vaccine Epitope Selection in PRRS

Reach 86.8 percent accuracy at 60 labels while outperforming random acquisition and nearing the 85 percent noise limit.

abstract click to expand

High-fidelity molecular docking simulations can produce biologically relevant estimates of epitope-receptor binding affinity but are computationally expensive and therefore limit the number of candidates that can be screened for vaccine design. In this work, we evaluate machine learning (ML) approaches where variants of active learning are used to classify instances of high binding affinity between 9-mer epitopes and a well-conserved swine leukocyte antigen (SLA) receptor in the context of Porcine Reproductive and Respiratory Syndrome (PRRS). We use an internally generated dataset of 80 epitope-SLA docking affinities, each requiring more than 48 hours of high-performance computing (HPC). Multiple model families (linear, MLP, CNN, and a small transformer) are trained under strict low-data conditions within a pool-based active learning loop. In each case, optimal model configurations are identified by conducting large-scale hyperparameter optimization over the combined space of model architecture, training configuration, acquisition policy, and ensemble decision rules. To mitigate the effects of data subsample selection, each candidate configuration is evaluated by averaging performance over many randomized and balanced training and validation data subsets. Across experiments, transformer-based sequence models consistently emerged as the best-performing architecture, with active incremental learning yielding significant improvement over a baseline random sample acquisition strategy. Under moderate training data availability (N=30), the optimized ML-model configuration outperforms a standard baseline trained on twice the amount of data. Under higher training data availability (N=60), the same configuration achieves a peak accuracy of 86.8%, consistent with an upper bound of 85% classification accuracy based on two independent estimates of conformational noise.

0

quant-ph 2026-06-29

High-entanglement ZZ map cuts overfitting in quantum binding classifier

by Aspen Erlandsson Brisebois, Luis Pablo Gonzalez Dominguez +9 more

Exploring the Effects of Entanglement on Quantum Machine Learning of Pathogen Epitope-Receptor Binding

All-to-all ZZ feature map delivers highest test-to-training AUAC ratio on 80-example PRRS epitope dataset while matching classical accuracy.

abstract click to expand

Parameterized quantum circuits (PQCs) provide a flexible substrate for hybrid quantum machine learning (QML), but their practical value on Noisy Intermediate-Scale Quantum (NISQ) devices remains an empirical question, especially because training depth and scale can introduce optimization challenges such as barren plateaus. Here we study how the number and topology of two-qubit entangling gates in the feature-map stage influence a fixed hybrid QNN workflow for classifying strong versus weak epitope-receptor binding in Porcine Reproductive and Respiratory Syndrome (PRRS) vaccine design. The dataset consists of docking-derived binding affinities for N=80 9-mer epitopes, labeled as Strong or Weak binding, and partitioned into training, validation, and test subsets using a 40:30:30 split. We compare a classical CNN benchmark with a hybrid Embedding-QNN architecture under four feature-map configurations: a non-entangling Z feature map, an all-to-all high-entanglement ZZ feature map, and two interleaved nearest-neighbour entanglement patterns of low and high depth. Among the configurations tested, the high-entanglement ZZ feature map is seen to provide the strongest evidence of reduced training-set overfit, with a lower training area under the accuracy curve (AUAC) and the highest test/training AUAC ratio, while preserving competitive test-set accuracy. These results do not establish a general QML advantage, but they suggest that feature-map entanglement topology is a meaningful design variable for sparse biological screening tasks and warrants further evaluation with additional metrics, larger datasets, and noise-aware or hardware-based experiments.

0

q-bio.PE 2026-06-29

Maximum-likelihood paths often atypical of real evolutionary histories

by Roberto Netti, Martin Weigt

Reconstructability of evolutionary intermediates in generative epistatic landscapes

Generative protein landscapes show conditional sampling recovers plausible ensembles better than point estimates, with topology setting the

abstract click to expand

Evolutionary intermediates connect observed proteins, but the sequence of steps that produced them is rarely recoverable from extant data alone. Here we ask what can, and cannot, be inferred about such intermediates from the endpoints. Using generative sequence landscapes as controlled models of protein-family evolution, we benchmark data-driven reconstruction against ground-truth simulated trajectories. We find that the best point prediction is not necessarily the most faithful evolutionary reconstruction: maximum-likelihood intermediates can be residue-wise accurate yet statistically atypical, whereas conditional sampling better captures the ensemble of plausible histories. Predictability is limited by the topology of the landscape. Constrained, low-mutability regions preserve information about the path, while permissive high-mutability regions open many alternative routes and erase path-specific memory. We also show that sequence divergence alone is an insufficient measure of elapsed evolutionary time; incorporating endpoint mutability provides a more reliable way to place intermediates in the landscape. These results recast intermediate reconstruction as a calibrated probabilistic problem. Rather than seeking a single "true" sequence, data-driven models should identify when endpoints contain evolutionary information, and return realistic ensembles.

0

q-bio.BM 2026-06-29

Method adds coevolution to ancestral protein reconstruction

by Alya Zeinaty, Leonardo di Bari +4 more

Towards coevolution-aware ancestral sequence reconstruction

DCA-integrated phylogeny produces ensembles that match both trees and natural sequence statistics under epistasis

abstract click to expand

Ancestral sequence reconstruction (ASR) is a powerful approach for studying molecular evolution and the emergence of protein function. Yet most ASR methods assume that sites evolve independently, neglecting the epistatic constraints that shape protein structure, stability, and function. This simplification affects both ancestral inference and its evaluation: maximum-a-posteriori reconstructions may over-concentrate probability into a single over-idealized sequence, whereas independent posterior sampling can generate implausible or poorly functional ancestors. Here, we introduce a coevolution-aware ASR framework that combines standard phylogenetic inference with Direct Coupling Analysis (DCA), thereby preserving site-wise ancestral uncertainty while enforcing residue-residue constraints learned from extant protein families. To benchmark the method, we develop a controlled forward-evolution framework based on a DCA evolutionary sampler, allowing reconstructed ancestors to be compared with known ground-truth sequences generated under realistic epistatic constraints. Applied to beta-lactamases and DNA-binding domains, the approach improves reconstruction when ancestral states are epistatically constrained, and yields ensembles of candidate ancestors that are both phylogenetically consistent and statistically compatible with natural protein families. This framework bridges the gap between single-sequence MAP reconstruction and unconstrained posterior sampling, providing a practical route toward ancestral reconstructions that better reflect the coupled nature of protein evolution.

0

cs.LG 2026-06-29

Two-stage tuning aligns proteins to target amino-acid mixes

by Violeta Basten-Romero, Rubén Muñoz-Tafalla +4 more

Two-Stage Fine-Tuning for Protein Sequence Generation with Targeted Amino-Acid Composition

Fine-tuning shifts average composition; reinforcement learning then enforces exact matches while sequence quality stays intact.

abstract click to expand

Protein language models are standard priors for biological sequence generation, but steering them toward explicit distributional design targets remains largely unexplored. We study a constrained protein generation problem in which sequences must match a desired amino-acid (AA) composition profile while preserving plausible sequence statistics and diversity. The motivating application is synthetic feed protein design, where the AA composition of dietary proteins directly determines their nutritional value. We propose a two-stage pipeline in which domain-adaptive fine-tuning (FT) on an in-domain protein dataset is followed by iterative reward-weighted FT via reinforcement learning (RL) anchored against the FT model as a frozen reference. We evaluate the pipeline on two AA compositions and find that FT brings the average composition close to the target, while the subsequent RL enforces specific sequence constraints that FT alone cannot satisfy. We additionally evaluate the design choices of the proposed composition reward term against two baselines and an ablated variant, isolate the contribution of each training stage, and verify that AA composition alignment is achieved without degrading sequence quality.

0

cond-mat.soft 2026-06-25

Vesicle solid domains roll, crumple then smooth with rising inflation

by Geunwoong Jeon, Anthony N. A. Prempeh +2 more

Frustrated shapes of solid domains in fluid membrane vesicles: From rolls and folds to crumples and wrinkles

Simulations show the sequence is set by two membrane tension scales and matches phase-separated lipid experiments

abstract click to expand

Fluid-solid composite vesicles, comprising 2D solid domains integrated into a topologically-closed fluid bilayer membrane, exhibit complex morphologies arising from the geometric frustration between spherical closure of the membrane and 2D solid elasticity. This scenario is distinct from the better studied case of multi-fluid domain vesicles. Here, we study the elastic energies and shape equilibria of a closed vesicle membrane containing a single, flexible circular solid domain using discrete finite-element (Surface Evolver) simulations, determining the key physical and mechanical parameters to govern shape selection. While we find that the 2D solid (shear) elasticity has minimal impact on the highly-under inflated morphologies, the geometrically non-linear resistance of the solid to Gaussian curvature substantially impacts the shape and elastic patterns form for inflated vesicles, by an amount that it grows with ratio of vesicle size to the elastic thickness of solid. For sufficiently large (thin) vesicles we characterize a generic sequence of ground state patterns of solid shape with increasing inflation: from cylindrical rolls and isometric folds to spatially complex patterns of crumples and wrinkles and ultimately to smooth caps. This sequence of non-isometric patterns at high-inflation is shown to be governed by the same far-from-threshold mechanics used to describe similar shape transitions in microscopic sheets on curved liquid interfaces, establishing that inflated shapes are governed by two basic mechanical scales of membrane tension. We find our predictions for highly-anisotropic shape equilibria of fluid-solid composite vesicles closely match experimentally observed shapes of giant unilamellar vesicles of phase-separated DPPC and DOPC.

0

q-bio.BM 2026-06-25

One decoder matches larger models across molecule design constraints

by Haoyu Lin, Yiyan Liao +4 more

Molexar: A Unified Multimodal Molecular Foundation Model for Drug Design

Embedding replacement during fine-tuning lets a small autoregressive model handle properties, proteins, and pockets while keeping 100 percen

abstract click to expand

Molecular generation is a central challenge in drug discovery, requiring models that explore vast chemical space while satisfying diverse design constraints. We present Molexar, a unified multimodal molecular foundation model built on Fragment-SELFIES, a robust, fragment-aware molecular language with validity-preserving decoding and explicit fragment structure. A pretrained autoregressive decoder learns the Fragment-SELFIES syntax and molecular distribution; supervised fine-tuning (SFT) then trains the same decoder on condition-molecule pairs spanning scalar molecular properties, pharmacophore fingerprints, protein sequences, and binding pockets, injecting each condition by in-place replacement of value-token embeddings so that all generation modes share one autoregressive path. Molexar achieves strong efficiency at a small parameter count while matching or exceeding larger models. The pretrained model reaches 100% validity and high drug-likeness in unconditional and fragment-constrained generation; the SFT model follows single- and multi-property instructions and remains competitive on target-conditioned generation on the CrossDocked2020 test set. On MolGenBench, Molexar further generates molecules with favorable safety and potency. These results establish Molexar as a practical unified foundation for computational chemistry and drug-design workflows.

0

q-bio.BM 2026-06-24

Command-line tool runs protein flexibility and docking simulations

by Chandran Nithin, Karol Wroblewski +6 more

CABS-flex standalone 3: an open command-line platform for protein flexibility simulation, peptide structure modeling, and protein-peptide docking

CABS-flex standalone 3 unifies coarse-grained modeling with all-atom reconstruction in an open Python package.

abstract click to expand

Summary: CABS-flex standalone 3 is an open command-line platform for fast CABS-based coarse-grained modeling of protein flexibility, peptide structures, and global or information-guided protein-peptide docking, coupled with all-atom reconstruction and analysis. The package builds on the established CABS-flex and CABS-dock ecosystem, widely used in structural bioinformatics for protein flexibility simulations and flexible protein-peptide docking. It provides a Python 3 implementation that brings together previous standalone functionality with recent developments in protein flexibility simulation, linear and cyclic peptide modeling, extended reporting and visualization, and deep-learning-based all-atom reconstruction with cg2all. Availability and Implementation: CABS-flex standalone 3 is implemented in Python 3 and is freely available as an open-source command-line package. Documentation is available at https://cabsflex.lcbio.pl. Source code is available at https://github.com/LCBio/CABSflex_standalone.

0

physics.bio-ph 2026-06-23

DNA hairpin sensor separates intercalators from groove binders

by Kiara Thompson, Kian Barnes +6 more

Electrochemical DNA Hairpin Sensors for Differentiating Small Molecule Intercalation from Minor Groove Binding

Four-base stem version matches known affinities and shows larger signal gains for intercalation than for minor groove binding.

abstract click to expand

Small molecule double-stranded DNA intercalators have significant potential for therapeutic applications. However, screening for and confirming a drug candidate's intercalative behavior remains labor-intensive and costly. To address this, we investigated the sequence and biophysical parameters that affect the performance of electrochemical DNA hairpin sensors for streamlined identification of structural intercalators. These sensors utilize oligonucleotide (oligo) sequences that form hairpins upon intercalator binding. The 3prime end of the oligo is modified with alkylthiol linkers for gold electrode surface monolayer self-assembly, while the 5prime end carries a methylene blue redox reporter. Hairpin formation enhances electron transfer between methylene blue and the gold electrode, which can be detected via voltammetry. We tested seven hairpin structures varying in stem length and sequence. Our optimal oligo, HP4, features a four-base-pair stem and responds to five DNA intercalators over a broad detection range, with EC50 in close agreement with published affinity (KD) values for these interactions. We further demonstrate HP4s ability to discriminate intercalator binding from a series of minor groove binders through significant differences in signal gain upon incubation. Altogether, our strategy establishes a platform for identifying intercalative compounds that should support the development of DNA-targeting therapeutics.

0

physics.chem-ph 2026-06-23

Eight-class alphabet turns cyclic peptide design into QUBO

by Yan Zhou, Yuqi Wang +1 more

Reduced-Alphabet QUBO/Ising Formulation for Constraint-Driven Cyclic Peptide Sequence Design

Binary variables over grouped amino acids encode ring-closure and composition rules for early-stage sequence search.

abstract click to expand

Cyclic peptide design requires balancing local residue preferences with constraints from ring-forming chemistry, residue spacing, topology, target compatibility, and developability. Here, we present a reduced-alphabet quadratic unconstrained binary optimization (QUBO)/Ising formulation for constraint-driven cyclic peptide sequence design. Amino acids are grouped into physicochemical or interaction-based residue classes, and peptide positions are represented by binary residue-class assignment variables. The objective combines one-hot sequence validity, cyclization constraints, optional target-compatibility terms, motif and composition rules, and coarse developability proxies. By modifying the relevant constraint terms, the same framework can represent head-to-tail, disulfide-bridged, stapled, and bicyclic peptide designs. A resource-aware eight-class alphabet motivated by MJ interaction-profile clustering is used as a default representation to balance coarse interaction-pattern preservation with encoding cost. The resulting QUBO/Ising objective is solver-agnostic and can be explored using classical or quantum-compatible binary optimization procedures. The model is intended as an early-stage search-space reduction and prioritization layer: it produces low-energy residue-class sequences rather than final molecular candidates, which require amino-acid decoding, cyclization-aware construction, and downstream structural or experimental validation.

0

q-bio.BM 2026-06-22

Model turns ligand 3D shapes into ready-to-make DNA libraries

by Zygimantas Jocys, Zhanxing Zhu +2 more

JEDEL: Zero-Shot DNA-Encoded Library Design for Early-Stage Drug Discovery

JEDEL builds focused libraries for 18 targets that score better on predicted binding than random selections, using only real building blocks

abstract click to expand

We present JEDEL, a framework for generating synthesis-ready DNA-encoded libraries (DELs) directly from three-dimensional pharmacophore representations of active ligands. JEDEL is the first model to map pharmacophore interaction patterns to actionable, scalable synthesis instructions, enabling the design of targeted libraries comprising potentially millions of molecules. Unlike existing generative approaches that produce virtual compounds requiring downstream synthesis planning, JEDEL operates within the space of purchasable building blocks and validated reactions, ensuring that every output is experimentally realizable by construction. JEDEL learns a predictive alignment between pharmacophore geometry and molecular structure and decodes this into combinatorial synthesis routes at scale. Across 18 protein targets, it generates focused libraries that outperform random and diversity-based baselines in predicted binding affinity, pharmacophore recovery, and sample efficiency, without target-specific retraining. JEDEL enables a shift from virtual molecule generation to experimentally deployable library design.

0

cs.CL 2026-06-22

One tokenization unifies bio sequences

by Qizhi Pei, Zhimeng Zhou +10 more

BioMatrix: Towards a Comprehensive Biological Foundation Model Spanning the Modality Matrix of Sequences, Structures, and Language

A decoder-only model maps molecules, proteins and language into shared tokens and trains with next-token prediction, matching specialized sy

abstract click to expand

We present BioMatrix, the first multimodal foundation model that natively integrates sequences, structures, and natural language for both molecules and proteins within a single decoder-only architecture. Existing biological foundation models pursue native multimodality and broad entity coverage separately: those that fuse multiple modalities under a shared objective remain confined to a single entity type, while those spanning multiple entity types either omit explicit structural modeling or rely on adapter-based designs in which the model cannot natively generate the very modalities it can read. BioMatrix closes this gap by mapping molecular sequences (supporting both SMILES and SELFIES notations), molecular structures, protein sequences, protein structures, and natural language into a shared discrete token space through a unified tokenization scheme, so that all modalities are consumed and produced uniformly under a single next-token prediction objective -- without external encoders, projection adapters, or modality-specific output heads. Built upon the Qwen3 language model (1.7B and 4B), BioMatrix is continually pretrained on 304.4 billion tokens spanning general and domain-specific text, sequence and structure views of molecules and proteins, and cross-modal corpora that interleave biomolecular entities with scientific text and link distinct entities through molecule-protein and protein-protein interaction data. After tuning on a comprehensive suite of downstream applications covering 80 tasks across 6 categories -- encompassing single-entity and multi-entity understanding and generation tasks across and within modalities -- BioMatrix achieves state-of-the-art or competitive performance on 77 out of 80 tasks, demonstrating that a single, natively multimodal generalist model can effectively match or surpass specialized approaches across a wide range of biological tasks.

0

physics.chem-ph 2026-06-17

Jeffreys measure normalizes Bayesian ensemble posteriors

by Ivan Gilardoni, Giovanni Bussi

Bayesian Sampling of Structural Ensembles: The Role of Ensemble-Counting Measures

Flat measure in Lagrange space produces non-normalizable distributions for finite trajectories; Jeffreys restores consistency and averages

abstract click to expand

Structural ensemble refinement is widely used to integrate molecular simulations with experimental measurements. While most applications focus on the maximum-a-posteriori (MAP) ensemble, Bayesian sampling of the posterior distribution can provide uncertainty estimates and posterior averages for arbitrary observables. A notable step in this direction was introduced by the Bayesian Energy Landscape Tilting (BELT) framework, where sampling is performed on a family of maximum-entropy ensembles parametrized by Lagrange multipliers. Here, we show that Bayesian sampling in this setting requires an explicit choice of ensemble-counting measure. In particular, the flat measure in Lagrange-multiplier space used in the original BELT formulation leads to a posterior distribution that is formally non-normalizable for finite reference trajectories. We propose the Jeffreys measure as an invariant ensemble-counting prescription, restoring normalizability in the finite-sample situations considered here, and providing a consistent definition of posterior averages. Using both an analytically tractable Gaussian model and maximum-entropy refinement of RNA oligomer simulations, we compare different ensemble-counting measures and show that they can significantly affect Bayesian estimates. The resulting methodology has been implemented in the \texttt{MDRefine} software package.

0

math.AT 2026-06-16

Knotting creates distinct scores in persistent homology cycles

by Aurelie Jodelle Kemme, Collins A. Agyingi +2 more

A Persistent Homology Signature of Knotting

Scoring hypergraph curvature on homology cycles reveals systematic differences in proteins and synthetic examples.

abstract click to expand

We ask whether knotting can be recognised using persistent homology. Starting from a point-cloud representation of a curve, we compute one-dimensional persistent homology, extract cycle representatives, and assign a hypergraph curvature-based score to these cycles. Motivated by proteins but tested more broadly, the method reveals systematic differences between knotted and unknotted structures in both protein families and synthetic examples. This suggests that knotting leaves a detectable persistent-homology-based signature.

0

q-bio.BM 2026-06-16

Thermodynamic hardware solves codon optimization at 10^6 lower energy

by Andraz Jelincic, Ross C. Walker

Energy-efficient codon optimization on thermodynamic hardware

The mapping to Ising sampling matches genetic algorithm scores on the SARS-CoV-2 spike protein.

abstract click to expand

The growing energy demand for computation is becoming increasingly unsustainable. Thermodynamic computing, which harnesses physical thermal fluctuations as a computational resource rather than suppressing them, offers orders-of-magnitude energy savings for probabilistic and combinatorial tasks. Pharmaceutical R&D, heavily reliant on computational optimization and sampling, is a natural application domain. Here we present what is, to our knowledge, the first concrete pharmaceutical application mapped to thermodynamic hardware with energy estimates grounded in prototype measurements. We reduce mRNA codon optimization, a combinatorial problem routinely solved in drug development, to sampling from an Ising model, making it directly executable on a thermodynamic sampling unit (TSU). Benchmarking three approaches (Potts sampling, Ising sampling, and a genetic algorithm baseline) on the SARS-CoV-2 spike protein, we find that all achieve comparable optimization quality (scores ~234-240), but energy estimates based on validated hardware models indicate that a TSU could solve this problem using approximately 10e6 times less energy than a conventional GPU. All code is released under an open-source license.

0

cs.LG 2026-06-15

Latent diffusion generates macrocyclic peptides with ring closures

by Junming Zhang, Siyu Yi +2 more

PepALD: Macrocyclic Peptide Generation via Autoregressive Latent Diffusion

The model embeds chemical structure, diffuses residues in latent space, predicts topology, and tunes for binding affinity.

abstract click to expand

Macrocyclic peptides are promising therapeutic candidates for intracellular targets, but their design requires simultaneous control over non-natural monomer chemistry, ring topology, membrane permeability, and target binding. Existing SMILES- or HELM-string generative models either operate in long atom-level sequence spaces or treat monomers as symbolic tokens with limited chemical grounding. We introduce PepALD, an Autoregressive Latent Diffusion (ALD) foundation model for \textit{de novo} macrocyclic peptide generation. The model represents HELM monomers with structured chemical embeddings, generates each residue through context-conditioned diffusion in chemically informed latent space, predicts R-group-aware ring closures during autoregressive generation, and aligns the denoiser to affinity rewards using winner-protected diffusion-adapted preference optimization. In silico experiments demonstrate PepALD's generation quality and reward-optimization performance against representative peptide generation baselines.

0

cs.AI 2026-06-12

Genomic profile sets Bayesian prior to separate nature from nurture

by Aruna Dey, Suraj Biswas

Is It You or Your Environment? A Bayesian Inference Framework for Genomically-Anchored Personalized Physiological Interpretation

Fixed genetic anchor distinguishes constitutional from environmental effects from the first measurement onward

abstract click to expand

Personalized health AI systems face a fundamental cold-start problem: machine learning models for physiological interpretation require weeks of individual behavioral data before they can distinguish constitutional variation from environmentally driven deviation. We propose a solution grounded in causal inference and Bayesian prior design. An individual's genomic profile serves as an exogenous genetic anchor -- a domain-informed, personalized prior that is fixed at conception, immune to reverse causation, and available before a single behavioral observation is collected. The anchor initializes a Bayesian belief state over an individual's physiological set point G-hat = mu + sum(beta_i * g_i), where beta_i are GWAS-derived effect sizes and g_i are risk-allele counts. Each incoming physiological measurement P produces a non-constitutional deviation delta = P - G-hat that separates the signal attributable to environment and state from the constitutionally fixed baseline. As behavioral data accrue, the prior decays according to G-hat_t = w(t)*G-hat_genomic + [1-w(t)]*P-bar_t, transitioning from genome-dominated to empirical-baseline-dominated inference. The same observed HRV of 55 ms generates a suppression hypothesis for a person whose prior predicts 80 ms, and an enhancement hypothesis for a person whose prior predicts 30 ms -- a reversal impossible without a personalized anchor. We develop this architecture across six physiological domains, grading genomic priors by evidence strength, distinguishing robustly replicated anchors (FTO, FADS1/2, FKBP5) from contested candidate genes (SLC6A4, MAOA, DRD2). We address the inference boundary between association, Mendelian randomization, and individual token causation, and define four constraints for deployment: evidence-graded priors, dynamic decay, ancestry-matched effect sizes, and attribution rather than deterministic output.

0

q-bio.BM 2026-06-12

Irregular focal-adhesion curvature keeps Piezo1 active in cancer cells under ultrasound

by Ivana Pajic-Lijakovic, Milan Milivojevic +2 more

Irregular curvature at focal adhesions modulates Piezo1 activity and low frequency ultrasound induced apoptosis in cancer cells

Regular curvature in healthy cells triggers cholesterol shifts that lower coordinated channel activity and block apoptosis.

abstract click to expand

Low-frequency, low intensity ultrasound (LIUS) has emerged as a promising physical modality capable of inducing selective apoptosis of cancer cells, while sparing healthy epithelial cells and fibroblasts. Hitherto, the mechanism underlying this selectivity has been unclear, but we now propose and develop a theoretical framework linking the distinct mechanical behaviours of cancer versus healthy cells to their differential responses to LIUS. We point out that cancer cells exhibit inhomogeneous ventral stress-fiber networks, which can produce irregular focal adhesion geometry and inward membrane curvature near focal adhesions under low-intensity ultrasound (LIUS). These curvature irregularities can favor loose packing of Piezo1 channels, thereby preserving their activity. In contrast, healthy epithelial cells and fibroblasts display more homogeneous cytoskeletal organization, which can result in more regular curvature profiles adjacent to focal adhesions. This leads to curvature-driven cholesterol redistribution, resulting in altered spatial organization of Piezo1 clusters and reduced coordinated channel activity and allowing cells to remain in their active, proliferative state when exposed to LIUS. Based on theoretical modeling and previous experimental findings, we propose that differences in cytoskeletal organization and membrane curvature can contribute to distinct Piezo1 activation patterns between healthy and cancerous cells. Our analysis identifies curvature-mediated Piezo1 redistribution as a potential physical basis for LIUS selectivity and provides a mechanistic foundation for designing ultrasound-based therapies to exploit the intrinsic cytoskeletal vulnerabilities of cancer cells.

0

cs.LG 2026-06-10

Multimodal student model matches teachers at lower cost

by Emily Nguyen, Yongchan Hong +3 more

GLACIER: A Multimodal Student-Teacher Foundation Model for Molecular Property Prediction

GLACIER fuses graph, SMILES, and descriptor encoders then distills from larger models to keep accuracy while cutting computation for propert

abstract click to expand

Deep learning models facilitate the discovery of molecules with tailored properties among billions of candidate compounds. However, the computational burden to develop and deploy state-of-the-art models continuously increases, limiting their scalability. Most large-scale models are unimodal in nature and overlook the potential to leverage complementary molecular data modalities. To address these shortcomings, this paper introduces the Graph-Language Alignment for Chemical Inference and Exploration using Representations (GLACIER) model, a student-teacher framework that integrates molecular graphs, SMILES strings, and physicochemical descriptors to learn rich molecular embeddings. Our framework consists of three stages: (1) we pretrain three student encoders on 100,000 drug-like molecules: a message-passing neural network for molecular graphs, a transformer-based encoder for SMILES strings, and a multilayer perceptron for physicochemical descriptors, (2) we fuse these student modalities using a novel Finsler geometry-aware module, and (3) distill complementary knowledge from large teacher models, including MiniMol and MolFormer, into a single lightweight model via contrastive learning. We demonstrate that GLACIER is a robust framework that delivers high predictive performance and computational efficiency in complex molecular property prediction tasks. Our code is publicly available at https://github.com/eemokey/glacier.

0

cs.LG 2026-06-10

Evolutionary kernels make sparse protein data sufficient for accurate models

by Martin Jankowiak, Yerdos Ordabayev +5 more

Flexible Kernels for Protein Property Prediction

Gaussian processes using substitution matrices and local linearity beat embedding baselines on binding and stability tasks.

abstract click to expand

Despite its importance to applications in protein design, predicting protein properties like binding affinity and thermostability from sparse experimental data remains a significant challenge. Accordingly, we introduce a class of sequence kernels that exploit evolutionary substitution matrices as well as local linearity and demonstrate that the resulting Gaussian processes provide data-efficient models of protein property landscapes, frequently outperforming alternatives that rely on foundation model embeddings. Furthermore--by learning what are in effect structure-aware substitution matrices--we show that our kernels can readily incorporate structural information from foundation models. We demonstrate that these structure-conditioned kernels are well suited to multi-task learning across multiple protein property landscapes and can decisively outperform local supervised learning methods.

0

q-bio.BM 2026-06-10

Helical structure key to modeling DNA shear rupture

by Ayman Hussein, Ralf Bundschuh

A kinetic model of shear-induced rupture of short dsDNA

A new master equation tracks single-base unzipping under force and matches experiments only when DNA's 3D twist is included in end-to-end ca

abstract click to expand

Force-induced dissociation of short double-stranded DNA (dsDNA) is central to single-molecule biophysics and DNA nanotechnology, yet a physically grounded kinetic description of shear-induced rupture for finite-length constructs remains lacking. Here we develop a master equation framework built on a force-dependent nucleation-zipper pathway with single-base transitions, enabling direct calculation of dissociation rates and transition state distances over a broad force range. Applied to a DNA-gold nanoparticle-DNA construct under constant shear force, the model accurately reproduces the experimental room-temperature data in the covered force regime and provides a unified interpretation of prior measurements on similarly sheared duplexes across all force regimes. A central result is that the three-dimensional helical geometry of dsDNA is essential for correctly defining the end to end distance under shear in the rod-like polymer model of short dsDNA. We further show that the extracted transition state distances are robust to variations in ssDNA polymer parameters within the experimentally relevant regime. Finally, we analyze the temperature dependence of the transition state distance and discuss how our framework captures globally-heated rupture while identifying the additional complications introduced by localized plasmonic heating in gold nanoparticle-coupled constructs. These results provide a predictive kinetic foundation for interpreting force-rupture experiments and for designing force- and temperature-actuated DNA nanostructures.

0

q-bio.BM 2026-06-08

Protein folding shifts from static structures to dynamic trajectories

by Juliette Griffié, Sviatlana Shashkova +41 more

Protein Dynamics Beyond Structure Prediction

Single-molecule data and multiscale models can now integrate to predict folding kinetics and assemblies in living cells.

abstract click to expand

The ability to predict protein three-dimensional structures from amino acid sequences is a landmark achievement in molecular biology, where recent deep learning approaches such as AlphaFold are the culmination of decades of work. Yet, the quantitative understanding of how protein sequences give rise to dynamic conformational changes and higher-order assemblies remains unsolved. Folding and conformational states are dynamic, stochastic processes, shaped by sequence, energy, co-translational constraints, chaperone machineries, and the physicochemical conditions of the cellular environment. Recent advances now position the field to move beyond static structural endpoints toward a mechanistic understanding of folding dynamics in living systems. Single-molecule techniques enable time-resolved observation of folding trajectories and intermediate states hitherto hidden by traditional structural biology approaches, while computational innovations and data-driven approaches offer new ways to integrate heterogeneous data across scales. In this Roadmap, we review the current conceptual landscape of protein folding, examine the experimental and theoretical gaps that remain, and discuss emerging strategies that integrate high-resolution measurements with multiscale modeling. We outline a roadmap toward a quantitative and predictive science of protein folding dynamics, conformational kinetics, and macromolecular self-assembly. Realizing this vision would transform our understanding of the dynamics of molecular self-organization, from the folding of individual polypeptides to the emergence of dynamic macromolecular complexes. This will enable rational control of folding and misfolding in health and disease, extend protein engineering principles beyond static structural design, and establish a mechanistic foundation for predictive and personalized interventions in proteostasis-related disorders.

0

cs.LG 2026-06-05

Generative models show weaker binding on shallow pockets

by Saket Reddy, Shiwei Liu

ShallowBench: Benchmarking Generative Drug Design Models on Shallow-Pocket Targets

New benchmark of 5780 low-concavity targets reveals state-of-the-art generative designs underperform on undruggable oncology sites.

abstract click to expand

While generative AI models have demonstrated remarkable success in structure-based drug design, they predominantly rely on deep binding pockets and struggle to sample effective ligands for challenging low-pocketability targets, such as the historically "undruggable" oncology targets KRAS and MYC. To address this gap, we introduce ShallowBench, a strictly curated benchmark of 5,780 shallow-pocket targets extracted from CrossDocked2020. By computing the difference between an Alpha Shape "lid" volume and the underlying protein atom voxel volume, we successfully isolated targets with low concavity while ensuring sufficient surface area for binding. Evaluating various state-of-the-art generative models reveals weaker predicted binding affinity on these low-concavity interfaces. ShallowBench therefore provides a rigorous benchmark for generative biology models and highlights the necessity of new architectural innovations or loss functions capable of navigating these challenging targets.

0

physics.chem-ph 2026-06-05

Three algorithms recover interaction potentials from XL-MS data in three-phase mixtures

by Börries von Seggern, Mohsen Sadeghi

Methods for Inferring Interaction Potentials from Cross-Linking Mass Spectrometry Data

Adapting inverse Henderson methods lets coordination numbers and similar observables determine pair potentials even when proteins form multi

abstract click to expand

Cross-linking mass spectrometry (XL-MS) has emerged as a powerful quantitative technique for probing intra-protein structural information as well as protein-protein interactions at an unprecedented scale. XL-MS data yield information on the pairwise spatial proximity of proteins through inter-molecular linkers. However, systematic methods for adapting such data for coarse-grained interacting particle models remain limited. Predominant focus is put on directly fitting radial distribution functions (RDFs), while numerous observables, e.g. coordination numbers, which are functionals of the RDF, cannot be uniquely inverted. In this work, we develop a framework for parameterizing interaction potentials from such observables in potentially phase-separated mixtures, as encountered in XL-MS results. We establish a connection between this problem and the inverse Henderson problem and adapt algorithms such as Iterative Boltzmann Inversion and Iterative Monte Carlo to its numerical solution. We derive exact and low-density limit gradient approximations and propose two new algorithms based on an adaptation of the predictor-corrector~framework. In total, we evaluate several optimization algorithms on biologically realistic ten-component test systems. We demonstrate that for homogeneous fluids, all methods achieve exceptional efficiency and accuracy. Critically, we further demonstrate successful parametrization in a challenging three-phase system. Here, three algorithms, namely Adam and gradient descent employing the low-density derivative as well as Newton's method with the exact gradient, reliably recover the correct parameters. These results establish a clear pathway from XL-MS experiments to coarse-grained protein models for systems where phase separation governs biological function, potentially enabling new investigations of biomolecular condensates and protein aggregation.

0

q-bio.BM 2026-06-04

Learned scorer produces conformation-selective protein binders

by Hanqun Cao, Zachary Quinn +5 more

AlloGen: Conformation-Selective Binder Generation with Differential State Scoring

It works across protein families and yields peptides that bind only the desired state in calmodulin experiments.

abstract click to expand

Protein binder design has largely optimized for affinity alone, leaving conformational selectivity unaddressed: for allosteric targets such as kinases, nuclear receptors, and GPCRs, a binder that engages both active and inactive states provides no functional specificity regardless of how tightly it binds. We introduce AlloGen, a modular framework that decouples backbone generation from a learned state-selectivity scorer $Q_\theta$, an SE(3)-invariant interface graph transformer trained via a two-phase curriculum that first learns interface geometry before imposing conformational discrimination. Because $Q_\theta$ is fully differentiable and generator-agnostic, it integrates with any backbone generator as a passive reranker or an active gradient-based guide without retraining. Across a diverse benchmark of proteins spanning multiple families and conformational mechanisms, AlloGen consistently identifies binders that preferentially recognize desired structural states while rejecting alternative conformations. Experimental validation on calmodulin further demonstrates that these computational selectivity signals translate to physical molecules, yielding de novo peptides that bind the desired holo conformation while exhibiting no detectable binding to the apo state. Together, these results establish conformational selectivity as a learnable property and provide a general framework for state-selective protein binder design.

0

q-bio.BM 2026-06-04

AF_Cache halves AlphaFold2 runtime on 5,000 protein pairs

by Sarah Narrowe, Arne Elofsson Claudio Mirabello

AF_Cache: Efficient Pipeline for Running AlphaFold for High-Throughput Protein-Protein Interaction Prediction

Feature caching and length bucketing deliver 2-fold inference speedup and 13-fold MSA acceleration for mitochondrial interactions.

abstract click to expand

Motivation: Accurate prediction of protein-protein interactions is essential for understanding biological processes, and recent advances such as AlphaFold2 and AlphaFold3 have enabled structure-based interaction prediction at unprecedented accuracy. However, the high computational cost of these methods, driven primarily by CPU-based repeated multiple sequence alignment (MSA) generation and, for AlphaFold2, repeated model recompilations, limits their applicability in large-scale, high-throughput settings. This creates a need for efficient pipelines that retain predictive performance while substantially reducing runtime. Results: We present AF_Cache, a high-throughput Nextflow pipeline for accelerating protein-protein interaction prediction using AlphaFold2 and AlphaFold3. AF_Cache combines GPU-accelerated MSA generation with MMseqs2, feature caching to eliminate redundant alignment computations, and sequence length bucketing to minimise repeated JAX compilations. Benchmarking on a dataset of 5,050 human mitochondrial protein pairs demonstrates a $\sim$2-fold reduction in inference time for AlphaFold2 and up to a 13-fold speedup of the MSA generation. AF\_Cache enables efficient large-scale interaction screening and provides a practical framework for deploying AlphaFold-based methods in high-throughput applications. Availability and implementation: The code and Nextflow pipeline are available on GitHub here: https://github.com/clami66/AF_cache. The code for reproducing the results of the paper, the MSAs, and the predicted models can be found at Zenodo: https://zenodo.org/records/20478892

0

q-bio.CB 2026-06-03

Stomatocyte volume alone sets spleen slit passage

by Zhaojie Chai, Jianlu Zheng +3 more

Quantifying the biophysical properties of stomatocytes in health and disease

Overhydrated cells need high pressure to pass while dehydrated ones thicken blood, separating filtration from viscosity effects

abstract click to expand

Hereditary stomatocytosis (HS) comprises red blood cell (RBC) disorders characterized by cup-shaped erythrocytes that respond oppositely to splenectomy: curative in overhydrated HS (OHS) but potentially thrombogenic in dehydrated HS (DHS/xerocytosis). This paradox persists because RBC biomechanics is governed by partly independent parameters--shear modulus, bending rigidity, surface-to-volume ratio (S/V), and cytoplasmic viscosity--that existing assays capture only piecemeal. Here we combine dissipative particle dynamics (DPD) simulations with microfluidic imaging to construct a control discocyte and three stomatocyte models (ST-RBC1-3) at fixed membrane area and decreasing volume (109.7, 101.5, 89.8 fL), spanning the OHS-to-DHS range. Tracing this parameter set through five mechanically orthogonal assays, we find that interendothelial-slit (IES) traversal is geometry-dominated: overhydrated ST-RBC1 requires an order of magnitude higher critical pressure than healthy RBCs, whereas dehydrated ST-RBC3 passes freely. ST-RBC3 nonetheless suppresses membrane tank-treading and raises low-shear whole-blood viscosity by ~29% at physiological haematocrit, comparable to Gaucher-disease hyperviscosity. A funnel-obstacle chip amplifies these differences into a label-free centerline-offset signal predicted to separate all four RBC types (~4.5 standard deviations between extreme phenotypes). These results unite single-cell mechanics, splenic filtration, and hemorheology in one framework, resolve the splenectomy paradox, and point toward microfluidic pre-operative risk stratification in HS.

0

q-bio.BM 2026-06-02

APLSuite runs CD4+ T cell epitope prediction in minutes via processing likelihood

by Jiarui Li, Marco K. Carbullido +3 more

APLSuite: An Integrated Suite for CD4+ T Cell Epitope Prediction via Antigen Processing Likelihood

The suite bundles APIs, Python tools, and a GUI so researchers finish APL calculations with GPU speed where earlier methods could not.

abstract click to expand

Computational epitope prediction is a critical tool for exploring and understanding CD4+ T cell-mediated immune responses, a key aspect of adaptive immunity. While existing computational methods primarily focus on supervised learning approaches, they often overlook the essential role of antigen processing in determining binding specificity. To address this limitation, our group developed Antigen Processing Likelihood (APL), an algorithm that integrates crystallographic B-factor, solvent accessible surface area (SASA), hydrogen exchange protection factors (COREX), and sequence entropy. In this paper we introduce APLSuite, a comprehensive and lightweight software suite designed to streamline APL-based epitope prediction. APLSuite integrates distributed RESTful API services, a Python client for data aggregation and processing, a data science tool for efficient epitope computation, and a user-friendly graphical user interface for non-coding users. It provides a seamless and efficient pipeline for APL calculation and epitope prediction that can be finished in minutes with GPU-acceleration, which has not been implemented by existed tools. This flexible and extensible software suite is deployable on desktop and cloud environments, offering both guided and customizable workflows to meet diverse research needs in immunology research and immunotherapy development. (The project page for this work is available at: https://tulane-mettu-landry-lab.github.io/blogs/APLSuite/)

0

q-bio.BM 2026-06-02

AI agent ranks protein sites for drug binding and suggests modalities

by Taehan Kim, Sarrah Rose Mikhail Leung +2 more

Site4Drug: Predicting Drug-Binding Target Sites with an AI Agent

Uses uniform evidence on topology, PTMs and sequence to rank sites and choose small-molecule or antibody approaches

abstract click to expand

Selecting where to intervene on a protein (i.e., choosing a targetable site) is often a more ambiguous and failure-prone bottleneck than selecting what binds, especially for membrane proteins where accessibility, topology, and post-translational modifications (PTMs) constrain actionable regions. We present Site4Drug, a modality-aware site-finding agent that outputs a ranked list of targetable regions with explicit constraints, evidence summaries, risk flags, and a traceable decision log. Rather than requiring users to specify the drug modality upfront, Site4Drug can recommend a binding modality (e.g., antibody/peptide-like vs small-molecule) from the same evidence used for site discovery, including topology, hydropathy, PTM propensity, disulfides, domain context, and sequence. Importantly, this evidence is applied consistently across modalities, including small-molecule pocket discovery, to avoid selecting chemically plausible but biologically occluded sites.

0

q-bio.BM 2026-06-02

Learned couplings improve biomolecular design validity

by Keyue Qiu, Xintong Wang +3 more

Demystifying Multimodal Biomolecular Co-design With Intrinsic Geodesic Coupling

Optimizing temporal alignment between sequence and structure generators reduces inconsistent states and yields more valid and diverse output

abstract click to expand

Biomolecules such as proteins and small-molecule ligands play a central role in biological systems, arising from the tight interplay between sequence and three-dimensional structure. Recent generative models for biomolecular co-design aim to capture this interplay by jointly modeling coupled modalities. However, existing approaches largely adopt a parallel execution of marginal generative processes, implicitly enforcing fixed synchronous coupling. We argue that a critical but overlooked degree of freedom lies in how these marginal processes are temporally coupled during training and generation, where inappropriate coupling can introduce high-variance supervision and inconsistent intermediate states, affecting modality consistency. To address this, we introduce GeoCoupling, a systematic framework that optimizes for temporal couplings between heterogeneous modalities. Empirical results across structure-based drug design and unconditional protein design demonstrate the learned couplings consistently outperform synchronous and randomly coupled baselines, yielding biomolecules with improved physical validity and diversity.

0

q-bio.BM 2026-06-02

Counter-diabatic term speeds peptide structure search on QAOA

by Sung Won Yun, Yeon Gyo Seo +4 more

Peptide Structure Prediction Using Counter-Diabatic Quantum Approximate Optimization Algorithm (CD-QAOA)

CD-QAOA produces heptapeptide lattice folds that match HF, DFT, MD and H-REMD results while converging faster than standard QAOA.

abstract click to expand

In this study, we predicted the structure of the heptapeptide APRLRFY, a neuropeptide sequence, on a tetrahedral lattice using a Quantum Approximate Optimization Algorithm (QAOA). QAOA is based on the adiabatic approximation and has been successfully applied to a wide range of optimization problems. However, relatively slow convergence during ground-state searches has frequently been reported. To overcome this limitation, we employed the Counter-Diabatic Quantum Approximate Optimization Algorithm (CD-QAOA), which introduces an additional counter-diabatic driving term into the adiabatic framework to accelerate convergence toward the ground state during peptide structure prediction. In the heptapeptide structure prediction, intermolecular interactions were modeled using two different approaches. In the first approach, only the interaction between the second residue, proline (P), and the seventh residue, tyrosine (Y), was included in the optimization. In the second approach, all residue-residue interactions within the heptapeptide were modeled using the Miyazawa-Jernigan (MJ) interaction matrix. To validate the peptide structures predicted using CD-QAOA, we additionally employed several classical computational methods, including quantum chemistry-based Hartree-Fock (HF) calculation and Density Functional Theory (DFT) calculation, conventional molecular dynamics (MD) simulation, and Hamiltonian replica exchange molecular dynamics (H-REMD) simulation. The structural similarities among the conformations obtained from these different approaches were systematically analyzed. CD-QAOA is highly effective for predicting the structures of short peptides. In particular, we demonstrate that a quantum-classical hybrid framework can significantly improve both the efficiency and accuracy of peptide structure prediction.

0

cs.LG 2026-06-01

Protein free energy reduces to unbalanced 2-to-1 norm

by Pratik Worah, Subhash Khot +1 more

Conditioned free-energy density of proteins using unbalanced solutions to constraint satisfaction problems

SDP algorithm solves the problem and locates flexible segments in Ubiquitin from its crystal structure

abstract click to expand

We show that computing the log-partition function (free-energy) of conditioned inhomogeneous Curie--Weiss spin Hamiltonians reduces to an unbalanced $2 \to 1$ norm computation, and design a polynomial-time SDP algorithm for this problem with a lower bound proof for the amount of unbalance achieved. Applied to the protein Ubiquitin, the framework starts from a known crystal structure, explores alternative backbone conformations across the free-energy landscape, and identifies flexible regions of the protein while preserving its native secondary structure.

0

cs.AI 2026-06-01

Controlled probes let LLM agents improve both affinity and druggability

by Zaifei Yang, Weiyu Chen +2 more

Probe Before You Edit: Probing-Guided Molecular Optimization for LLM Agents in Structure-Based Drug Design

Diagnostics reveal current pipelines rarely achieve joint gains; PROBE maps probe responses to guide edits and reaches SOTA on CrossDocked20

abstract click to expand

Structure-based drug design increasingly employs LLM agents to iteratively refine ligands against a target pocket, yet a viable ligand must satisfy two often-conflicting objectives -- binding affinity and druggability -- which single optimization steps rarely improve together. To quantify this difficulty, we introduce two diagnostic metrics: the first measures how often a single edit improves both objectives, and the second measures how often a gain on one objective comes with a loss on the other. Applying these diagnostics to current LLM-agent pipelines exposes a consistent failure mode: the agent performs molecular editing without knowing how the pocket-ligand complex responds to local modifications, thus rarely achieving joint improvement. Inspired by medicinal chemists, who probe the pocket-ligand complex with controlled analog edits before choosing an optimization direction, we propose \textbf{PROBE}, an optimization framework built around edit-response probing. PROBE first decomposes the ligand into editable sites and builds a pocket-specific \textbf{site map} that flags where joint gains are plausible, where the two objectives are likely in tension, and where liability substructures should be changed; it then performs controlled probe edits whose responses are distilled into an \textbf{EditManual}. Guided by the site map and EditManual, PROBE runs an iterative multi-agent loop in which an affinity agent, a druggability agent, and a co-optimization agent jointly produce edits. On the CrossDocked2020 benchmark, PROBE achieves state-of-the-art performance and substantially mitigates the failure modes exposed by our diagnostics metrics.

0

cs.LG 2026-06-01

Energy-based surrogates let flow models anneal molecular samples without divergence cost

by Daniel Peñaherrera, Rishal Aggarwal +1 more

Scalable Inference-Time Annealing with Surrogate Likelihood Estimators

SITA achieves state-of-the-art sampling on alanine dipeptide and tripeptide by swapping expensive score divergences for fast surrogate likel

abstract click to expand

A long standing challenge in computational chemistry and biophysics is efficiently sampling the Boltzmann distribution of molecules. Advances in generative modeling have been proposed to address the limitations of conventional sampling techniques by eliminating the computational cost of simulation. A promising direction is iteratively finetuning diffusion models along a temperature ladder whereby training data is generated via importance sampling during inference-time annealing. Unfortunately, these methods require computing a divergence over the score field to estimate importance weights, rendering them intractable for larger systems. Here we present scalable inference-time annealing (SITA), which retrains flow-based models to generate samples at progressively lower temperatures using an energy-based model to facilitate fast surrogate likelihoods. We demonstrate state-of-the-art performance on both Alanine Dipeptide and Alanine Tripeptide while avoiding costly divergence terms. Our code is available at https://github.com/countrsignal/sita.git

0

q-bio.BM 2026-06-01

Zero-shot mRNAs reach 400-fold higher expression than wild-type

by Sawan Patel, Sophia Tang +9 more

mRNAutilus: Multi-Objective-Guided Discrete Generation of mRNA with Optimized Therapeutic Properties

Single diffusion process with multi-objective guidance designs full transcripts for luciferase, Spike, and editing tools without separate UT

abstract click to expand

Therapeutic mRNA design requires coordinating multiple interacting sequence features across the full transcript, where codon usage, untranslated regions (UTRs), and their coupling jointly determine stability, translation efficiency, and protein expression. Here, we present mRNA generation via unrolled trajectories and informed latent updates (mRNAutilus), a framework for simultaneous codon optimization and de novo UTR design directly from sequence. mRNAutilus combines a masked discrete diffusion model trained on millions of full-length mRNAs with Monte Carlo Tree Guidance to generate Pareto-efficient sequences under multiple functional objectives, using lightweight regressors over model embeddings to predict half-life, translation efficiency, and protein abundance. Unlike recent methods that design coding sequences and UTRs separately or rely on post hoc assembly and screening, mRNAutilus generates complete transcripts in a single process optimized across properties. Across diverse targets, zero-shot mRNAs encoding P. pyralis luciferase achieve over 400-fold higher expression than wild-type and outperform commercial and machine learning-designed baselines, including zero-shot generative approaches. Zero-shot SARS-CoV-2 Spike mRNAs exceed clinically used and commercial constructs and match or surpass lab-optimized designs with improved durability. We further demonstrate generality in therapeutic settings, including prime editing (PEMax) and programmable proteome modulation, where mRNAutilus-designed constructs enhance expression of peptide-guided E3 ligases (uAbs) for beta-catenin degradation. These results establish a sequence-based, multi-objective framework for generating functional mRNAs tailored to diverse biological applications.

0

q-bio.BM 2026-06-01

Framework designs fluorescent biosensors for any small molecule

by Bowen Jing, Mihir Bafna +6 more

SwitchCraft: A Programmatic Framework for Designing State-Switching Proteins

Backpropagation through structure prediction models creates proteins that switch states in response to chosen analytes.

abstract click to expand

Multistate mechanisms underlie many of the complex functions observed in natural proteins. The ability to rationally design multistate proteins would have transformative implications for many areas of biotechnology, yet lies beyond the capabilities of existing deep learning frameworks for protein design. To address this gap, we introduce SwitchCraft, a versatile and programmatic framework for designing state-switching proteins based on backpropagation through compositional design constraints parameterized by structure prediction models. In silico evaluations demonstrate success on a wide range of state-switching functional primitives, from allosteric regulation of motifs to discrimination of bound ligand identities. Using these primitives, we demonstrate an in silico strategy for de novo design of fluorescent biosensors to arbitrary small molecule analytes. These results position SwitchCraft at the inception of a powerful paradigm for higher-order functional protein design. Code is available at https://github.com/bjing2016/switchcraft.

0

q-bio.BM 2026-06-01

One model unifies protein understanding and design in LLMs

by Keyue Qiu, Yixin Wu +20 more

AMix-2: Establishing Protein as a Native Modality in Large Language Models

Shared tokens and block diffusion let AMix-2 handle both tasks and beat general language models on a new realistic benchmark.

abstract click to expand

We present AMix-2, a protein-text foundation model that establishes protein as a native modality in large language models (LLMs), unifying protein understanding and sequence design within a single foundation model. AMix-2 is built upon two key ideas: (1) a unified protein-text formulation that embeds natural language and protein sequence in a shared token space, enabling one model to perform biological reasoning and conditional design instead of separate downstream task-specialized models; and (2) a block-wise diffusion language modeling backbone that combines causal generation across blocks with bidirectional context and iterative refinement within blocks. This scheme better matches the intrinsic nature of proteins than a strict left-to-right factorization. To evaluate protein foundation models under realistic generalization settings, we further introduce ProteinArena, a comprehensive benchmark with time-aware and homology-aware protocols across various understanding and design tasks, and with baselines covering classical bioinformatics tools, protein-specialized models and LLMs. On ProteinArena, AMix-2 outperforms frontier LLMs and demonstrates competitive performance to task-specific protein models. Controlled experiments further show that the diffusion-based paradigm generally surpasses its autoregressive counterpart, highlighting the advantage of flexible generation order for protein sequences. We release both AMix-2 and ProteinArena to facilitate open research in protein foundation models.

0

cs.LG 2026-05-28

Residue-level MaxSim improves remote protein homolog search

by Gabrielle Cohn, Rohan Gumaste +2 more

PROTOCOL: Late Interaction Retrieval for Protein Homolog Search

Late interaction on language-model embeddings beats pooled vectors and alignments on SCOPe superfamily and Pfam clan benchmarks.

abstract click to expand

Protein homology search underlies function annotation, structure prediction, and evolutionary analysis, but remains challenging in the "twilight zone," where global sequence similarity is weak and classical alignment methods lose sensitivity. Protein language models provide context-aware representations that could improve alignment sensitivity in this regime. However, prior protein embedding-based retrieval pipelines often pool these representations into a single vector, potentially obscuring local motifs, domains, or conserved residues that reveal remote homology. We introduce ProtoCol, a model which represents proteins as sets of residue embeddings and uses ColBERT-style late interaction to test whether residue-level comparison improves homolog retrieval. ProtoCol encodes proteins independently, keeps candidate representations pre-computable, and scores candidates with MaxSim over residue embeddings. On SCOPe superfamily and Pfam clan benchmarks, ProtoCol outperforms sequence-composition, alignment-based, pooled PLM, and trained single-vector baselines, supporting late interaction as an effective retrieval layer for remote homology search.

0

physics.bio-ph 2026-05-28

Virological data are protocol kernels

by Lillian St. Kleess

Experimental Collapse in Virophysics: Protocol-Resolved Observation, Inference, and Plaque-Assay Blindness

The plaque assay measures effective infectious concentration under its specific conditions rather than total particle numbers.

abstract click to expand

Virological measurements are often treated as reports of virion structure, mechanics, dielectric response, infectivity, or titer. In practice, an experiment observes a protocol-conditioned projection of a richer latent virion--environment ensemble. This paper defines this process as experimental collapse within protocol-resolved virophysics. Its central object is the null-inclusive observation operator $P_{\mathrm{obs},t}^{\varnothing}(\,\cdot\mid E\,) = \mathcal{M}_{E,t}^{\varnothing}P_{\mathrm{ref},t}$, which maps a reference latent ensemble to the observed ensemble generated by protocol $E$, including null outcomes. The formulation separates latent-state transformation, detection weighting, readout, and non-observation, making protocol effects explicit components rather than bias terms. The framework introduces protocol-conditioned latent ensembles, collapse functionals, protocol blindness, observation equivalence, Fisher-information observability, inverse inference, and multi-protocol consistency. It identifies collapse mechanisms including preparation, surface immobilization, mechanical loading, field steering, medium filtering, amplification, censoring, and detection thresholds. As a worked example, the plaque assay estimates an effective protocol-conditioned infectious concentration $\Lambda_{\mathrm{PFU}}=\int_{\Psi}\pi_{\mathrm{PFU}}(x;E_{\mathrm{PFU}})n_{\mathrm{ref}}(x),dx$, rather than total particle concentration. This recovers the Poisson plaque-count model and PFU titer formula in the dilute regime; extensions to overdispersion, zero inflation, plaque merging, endpoint dilution, neutralization, and morphology-augmented readouts recast deviations as protocol-conditioned information. Thus, virological data are outputs of explicit protocol kernels, clarifying what measurements report, miss, and how complementary assays can infer hidden latent virion structures.

0

q-bio.BM 2026-05-26

Surface manifold model outperforms backbone methods on protein binder and enzyme design

by Fang Wu, Shuting Jin +6 more

SurfDesign: Effective Protein Design on Molecular Surfaces

SurfDesign uses continuous geometric representations and equivariant passing to improve functional de novo designs over prior approaches.

abstract click to expand

Protein function is largely determined by molecular surface geometry and physicochemical complementarity, yet most protein design methods condition only on backbone structure. We introduce SurfDesign, a surface-conditioned protein design framework that models molecular surfaces as continuous geometric manifolds and integrates them with pretrained protein language models. SurfDesign employs surface-based equivariant message passing to capture surface normals, curvature, and directional geometry, together with a parameter-efficient fine-tuning strategy. Focusing on functional protein design, we show that SurfDesign consistently outperforms prior surface-conditioned and backbone-only methods on de novo binder and enzyme design benchmarks. We also report strong performance on inverse-folding benchmarks as a diagnostic of structural compatibility. Our results highlight manifold-aware surface representations as a principled foundation for functional protein and enzyme design. Code is available at https://github.com/smiles724/SurfDesign.

0

q-bio.BM 2026-05-26

Montparnasse solves RNA puzzles over three times faster

by Tristan Cazenave

The Montparnasse Algorithm for RNA Design

The Monte Carlo method reaches full coverage of the Eterna100 benchmark quicker than prior approaches and improves mRNA designs.

abstract click to expand

RNA design consists of discovering a nucleotide sequence that optimizes predefined criteria, such as secondary structure. It is useful for synthetic biology, medicine, and nanotechnology. We propose Montparnasse, a Monte Carlo search framework based on Generalized Nested Rollout Policy Adaptation, augmented with a problem-specific prior, slow and long adaptation at level 1, and a lexicographic multicriteria evaluation. Montparnasse solves all 100 puzzles of the Eterna100 V1 benchmark consistently faster than DesiRNA, the previous state of the art, across all time limits, reaching full coverage more than three times faster overall. On messenger RNA secondary structure optimization for hemoglobin alpha, it identifies sequences with more paired bases than the MFE-optimal solution of LinearDesign.

0

cs.LG 2026-05-26

Mass spec data steers diffusion models to better protein complexes

by Alon Shtrikman, Nitzan Simchi +4 more

Co-folding model guided by structural proteomics

AIMS-Fold converts XL-MS and HDX-MS measurements into potentials that guide sampling and beat unguided models on induced proximity targets

abstract click to expand

Protein structure generative models excel at predicting single protein static structures from sequence, but routinely fail to capture the correct conformational state of protein complexes, critical for protein design and induced proximity modalities such as antibodies and PROTACs. While structural proteomics techniques like Cross-Linking Mass Spectrometry (XL-MS) and Hydrogen-Deuterium Exchange (HDX-MS) offer valuable spatial and dynamic insights, integrating these sparse, heterogeneous measurements into these models remains an open challenge. Here, we bridge this gap by combining structural proteomics data with the rich biophysical priors learned by pretrained diffusion models. We introduce AIMS-Fold, an inference-time guided-diffusion framework that actively steers the generative sampling trajectory using differentiable physical potentials derived from XL-MS spatial restraints and HDX-MS solvent accessibility profiles. We demonstrate that these structural methods individually enhance predictive accuracy, and their integration yields synergistic improvement. Crucially, by leveraging these experimental restraints, AIMS-Fold achieves higher accuracy on challenging induced proximity targets than purely computational, unguided state-of-the-art models like Boltz-2. This establishes our framework as a powerful, integrative computational approach for the structure based drug design of induced proximity drugs. Evaluation code will be made publicly available upon publication.

0

cs.DB 2026-05-25

Knowledge graph preserves metabolomics provenance links

by Matthieu Féraud, Dina Boukhajou +2 more

MetaboKG: An Analysis-centric Knowledge Graph Framework for Untargeted Metabolomics

Workflow and extended identifiers turn scattered spectra and annotations into traceable, queryable data.

abstract click to expand

Untargeted metabolomics generates large volumes of tandem mass spectrometry (MS/MS) data and computational annotations that can reveal molecular mechanisms across organisms and environments. Public reuse has improved through harmonized repository metadata and access infrastructures such as Pan-ReDU, and through metabolomics knowledge graphs such as ENPKG and METRIN-KG. Yet the analytical layer remains fragmented: spectra, features, workflow outputs, annotations, confidence evidence, and contextual metadata are still scattered across repositories and tabular artifacts. We present MetaboKG, an analysis-centric knowledge graph framework for engineering reusable metabolomics knowledge from public repositories, metadata, and GNPS molecular network results. MetaboKG contributes a transformation workflow that preserves links between repository exports, analytical files, spectra, features, and annotation results; a semantic model grounded in PROV-O and SIO and aligned with the Mass Spectrometry ontology (MS), ChEBI, NCBITaxon, ENVO, and NCIT to represent provenance, analytical evidence, metadata attributes, and controlled vocabulary terms; and a Universal Annotation Identifier strategy extending the Universal Spectrum Identifier (USI) with workflow-specific components for late binding, incremental ingestion, and post hoc linkage across analyses. We demonstrate MetaboKG at the public-repository scale on 680 GNPS molecular networking results and evaluate it through competency questions covering biochemical enrichment, environmental specificity, and cross instrument analytical variation. Results show that graph-based integration supports traceable annotation reuse and reproducible SPARQL exploration of biochemical relationships that remain fragmented across repository-native resources.

0

cs.AI 2026-05-25

Text from enzymes improves bidirectional reaction retrieval

by Yuhang Zhang, Keyan Ding +6 more

TIGER: Text-Informed Generalized Enzyme-Reaction Retrieval

TIGER fuses generated descriptions with sequences for consistent mapping across datasets and directions.

abstract click to expand

Enzyme-reaction retrieval is a fundamental problem in computational biology, underpinning enzyme characterization, reaction mechanism elucidation, and the rational design of metabolic pathways and biocatalysts. As a bidirectional task, it entails both enzyme-to-reaction and reaction-to-enzyme mapping. However, existing approaches suffer from poor generalization across tasks and distributions, with performance highly sensitive to dataset splits and substantial asymmetry between retrieval directions. To address these challenges, we present TIGER, a Text-Informed Generalized Enzyme-Reaction Retrieval framework that leverages protein-to-text generation models to distill textual semantic knowledge from enzyme sequences, providing a generalized representation that bridges enzymes and biochemical reactions. To ensure the quality and reliability of textual semantics, we design a Dynamic Gating Network that adaptively fuses text-derived knowledge with sequence features, enabling more consistent and informative enzyme representations, while a Structure-Shared Feature Projector aligns enzyme and reaction representations within a unified latent space. Extensive experiments demonstrate that, under bidirectional retrieval supervision, TIGER significantly outperforms state-of-the-art baselines across diverse distributions and exhibits strong robustness and transferability across tasks.

0

q-bio.BM 2026-05-22

Three-view pretraining lifts protein structure tasks

by Taewon Kim, Hyosoon Jang +7 more

Atom-level Protein Representation Learning Improves Protein Structure Prediction

TriProRep outperforms sequence and prior models on homodimer co-folding and interaction prediction by recovering tokens across amino-acid, b

abstract click to expand

Recent advances in generative modeling show that pretrained representations can improve generation as conditioning features or alignment targets. Motivated by this, we study protein representations for predicting structures beyond conventional function annotation. We propose TriProRep, a structure-aware pretraining method that jointly models three aligned residue-level views: amino-acid identity, backbone geometry, and local full-atom geometry, discretely encoded via VQ-VAE tokenizers. By pretraining to recover original tokens from generator-corrupted views, TriProRep learns to distinguish plausible but incorrect cross-view augmentations from the original protein. We further introduce RepSP, a benchmark for evaluating protein representations in structure-predictive settings. RepSP tests three uses of representations: homodimer co-folding from apo-chain representations, residue-level prediction of homodimer-derived interaction properties, and representation-aligned monomer structure prediction. Across these tasks, TriProRep improves over sequence-only and prior structure-aware representation models, while maintaining competitive performance on conventional benchmarks.

0

q-bio.BM 2026-05-20 2 theorems

Proteomes share fixed elemental ratios from LUCA to viruses

by L. Felipe Benites, Louie Slocombe +1 more

Deep-time consistency in proteome elemental composition across cellular and viral life

Modern life and ancient ancestors fit tight carbon-hydrogen-nitrogen-oxygen-sulfur limits that reduced primordial alphabets violate.

abstract click to expand

Proteins are constructed from a limited alphabet of ~20 amino acids, yet the origins and selection of this specific alphabet are unresolved. One largely overlooked aspect is whether elemental composition constrains the range of viable proteomes. Here, we analyze the elemental composition of thousands of proteomes spanning cellular domains and viral realms. Despite evolutionary divergence and orders-of-magnitude variation in proteome size and gene content, proteomes exhibit strikingly consistent elemental composition. This consistency is substantially more constrained than amino acid frequencies or physicochemical properties and is not explained by evolutionary relatedness, biological function, or amino acid usage alone. Viral proteomes occupy the same elemental composition space observed in cellular organisms despite the absence of a single viral common ancestor, suggesting common biochemical constraints shape proteome organization across life. To investigate the evolutionary origins of this pattern, we compare modern proteomes with multiple independent reconstructions of the Last Universal Common Ancestor (LUCA) and with synthetic reduced-alphabet proteomes generated from primordial amino acid alphabets. LUCA proteomes occupy the same constrained elemental composition space observed in modern Bacteria and Archaea, whereas reduced primordial-like alphabets systematically generated alternative elemental regimes outside the modern range despite retaining high sequence similarity to extant proteins. Reduced alphabets disrupt fold space and reorganize relationships between elemental composition and predicted protein structural organization. Our results suggest that constrained elemental composition represents a fundamental organizational property of proteomes, which emerged early in evolution and may have contributed to the selection and stabilization of the modern amino acid alphabet.

0

q-bio.BM 2026-05-20 2 theorems

Microbial chemistry clusters in distinct elemental space

by Pilar C. Vergeli, Cole Mathis +6 more

Elemental Stoichiometry as an Ecological Biosignature with Applications to Life Detection

Enrichment in heteroatoms and shifted ratios set biological samples apart from synthetic and planetary data, offering a statistical test for

abstract click to expand

The vast chemical space of possible small molecules, estimated at 10^60 compounds for molecules composed of just C, N, O, and S, is only sparsely occupied by biology. We propose that where life selects molecules within this space constitutes a detectable ecological signature: a fingerprint not of specific compounds, but of the statistical structure of elemental composition across molecules sam-pled from ecological systems. Here we introduce a framework combining Van Krevelen diagrams and element scaling laws to characterize the elemental composition of regions of chemical space occupied by biological systems and contrast them with other chemical systems. Applying this framework to 11,834 microbial metagenomic samples, we show that microbial metabolisms occupy a region of chemical space, which is enriched in heteroatoms such as P, S, N, and O relative to C, shifted toward higher O:C and H:C ratios. We observe sublinear element scaling with system size, yielding insights into how elemental constraints dictate how biological systems occupy chemical space. These patterns are distinct from a sample of 18,000 compounds from the comprehensive Reaxys synthetic chemical database. Critically, datasets from molecules detected in planetary science mission data occupy statistically distinct regions from both terrestrial biological and Reaxys distributions, demonstrating that with standardized methods for data collection, the approach could be developed to discriminate biotic from abiotic chemical signatures in small molecule data from planetary science missions. Our work shows how a combination of Van Krevelen fingerprinting and elemental scaling laws can provide a new class of ecological biosignatures for life detection leveraging mass spectrometric data from planetary missions, which could generalize beyond Earth's specific biochemistry.

0

cs.LG 2026-05-19 Recognition

Self-supervised autoencoders beat supervised ones on protein fold tasks

by Dexiong Chen, Andrei Manolache +2 more

Protein Fold Classification at Scale: Benchmarking and Pretraining

A new non-redundant benchmark shows high-masking MiAE models scale better than large supervised networks for recognizing protein topologies.

abstract click to expand

Classifying protein topology is essential for deciphering biological function, but progress is held back by the lack of large-scale benchmarks that avoid duplicates and by models that do not scale well. We introduce TEDBench, a large-scale, non-redundant benchmark for protein fold classification constructed from the Encyclopedia of Domains (TED) and Foldseek-clustered AlphaFold structures. We show that on TEDBench, current protein representation learning methods either require very large models or fail to deliver strong performance. To address this challenge, we propose Masked Invariant Autoencoders (MiAE), a self-supervised framework for protein structure representation learning. MiAE uses an extremely high masking ratio of up to 90% with an $\mathrm{SE(3)}$-invariant encoder and a lightweight decoder that reconstructs backbone coordinates from the latent representation and mask tokens. MiAE scales well and outperforms supervised counterparts and state-of-the-art baselines on TEDBench, establishing a strong recipe for protein fold classification. To test transfer beyond AlphaFold structures, we further benchmark on a curated dataset from experimental structures of CATH v4.4. TEDBench is available at https://github.com/BorgwardtLab/TEDBench.

1 0

0

q-bio.BM 2026-05-18 2 theorems

MoleCode gives LLMs direct structural access to molecules

by Zhiyuan Yan, Chen Liu +9 more

MoleCode unlocks structural intelligence in large language models

Explicit Subgraph-Node-Edge grammar replaces hidden topology in SMILES, shortening reasoning and enabling precise edits.

abstract click to expand

Molecules are graphs, but large language models~(LLMs) are usually asked to reason about them through linear strings. The most popular molecular representation, SMILES, compresses atoms, bonds, branches and rings into a compact sequence in which topology is implicit, forcing LLMs to reconstruct molecular structure before performing the requested chemical operation. Here we introduce MoleCode, an LLM-native, training-free, graph-explicit molecular language in which all molecular components are represented as typed entities with persistent identifiers and explicit relations. MoleCode makes molecular topology directly readable, editable and auditable within the language context, allowing an LLM to operate on structure rather than recover it from syntax. Across molecular reasoning, editing, generation and analysis tasks, this representational shift improves frontier LLMs most strongly when structural access is limiting: unfamiliar molecules, topology-sensitive operations, larger structures and repetitive polymers. It also changes how inference is allocated, replacing long reasoning traces devoted to implicit structural reconstruction with shorter, more chemically directed reasoning over explicit atoms and bonds. In molecular optimization, this enables localized, property-aligned edits that preserve structural similarity to the starting compounds. The same Subgraph--Node--Edge grammar extends beyond small molecules to polymers, Markush structures, mechanism-style transformations and interleaved scientific documents, including research articles and patent disclosures in which chemical information is distributed across text and images. These results suggest that the interface between scientific objects and LLMs should not treat structure as something to be decoded from text. When the object of reasoning is relational, the structure itself should be part of the language.

0

q-bio.BM 2026-05-18

Ligand diffusion raises protein TM-score from 0.67 to 0.80

by Chen Wei, Fanding Xu +6 more

Ligand-Conditioned Discrete Diffusion for Protein Sequence-Structure Co-Design

Token-space model conditions discrete diffusion on small-molecule geometry, cutting pocket RMSD and lifting docking success rates.

abstract click to expand

Proteins perform their biological functions through three-dimensional structures encoded by amino acid sequences, and ligand-binding protein co-design requires models that generate sequence-structure compatible proteins under explicit ligand constraints. Although continuous diffusion and flow-based models support ligand-aware design in coordinate or latent spaces, existing discrete diffusion protein language models mainly operate over sequence or structure tokens without direct small-molecule conditioning. We introduce \textbf{ProtLiD$^2$}, a \textbf{Prot}ein \textbf{L}igand-conditioned \textbf{D}iscrete \textbf{D}iffusion model for protein sequence-structure co-design. ProtLiD$^2$ jointly generates amino-acid sequence and discrete structure tokens while incorporating ligand chemical and geometric information through geometry-aware cross-attention. Trained on over one million ligand-protein complexes, ProtLiD$^2$ extends masked discrete diffusion to ligand-aware functional protein design. We further propose maximum confidence-margin guided ReMask decoding, an inference-time self-correction strategy that retains confident predictions and remasks uncertain tokens. ProtLiD$^2$ improves global fold confidence over Complexa in whole-protein design, increasing TM-score from 0.672 to 0.802 and pLDDT from 64.55 to 73.00. In pocket co-design, ProtLiD$^2$ reduces active-site BB-RMSD from 3.46/3.40{\AA} for FAIR/PocketGen to 1.97{\AA}, and improves ligand-aware pass rates over PocketGen from 14.86% to 59.73% and from 6.08% to 23.49% under stricter docking thresholds. These results support ligand-conditioned discrete diffusion as an effective token-space framework for functional protein co-design. Code will be available at https://github.com/auroua/ProtLiD.

0

q-bio.BM 2026-05-15 Recognition

NMR simulations quantify entropy shifts from Fyn SH3 mutations

by Kresten Lindorff-Larsen, Robert B. Best +4 more

Detection of residual native state entropy changes upon mutation in Fyn SH3

Restrained dynamics show mutation effects equal several kcal/mol in free energy and contribute to stability changes

abstract click to expand

NMR relaxation experiments have shown that there are small but measurable changes in the native state dynamics of the Fyn SH3 domain associated with the substitution by other amino acids of a phenylalanine residue (F20) in the hydrophobic core. We have here used experimental values of NMR order parameters for the wild type protein and two mutational variants (F20L and F20V) as restraints in molecular dynamics simulations. This approach is highly sensitive and provides an atomistic description of the subtle perturbations in native state fluctuations accompanying the mutations. The structural ensembles that we have determined using this method allow the changes in the native state entropy of the protein caused by each of the mutations to be estimated. These entropy changes correspond to free energy variations of several kcal/mol and therefore represent sizable contributions to the overall changes in stability that are associated with the amino acid mutations.

0

cs.LG 2026-05-14 Recognition

New tokens encode protein motion ensembles

by Kaiwen Shi, Carlos Oliver

ENSEMBITS: an alphabet of protein conformational ensembles

Ensembits derives discrete symbols for conformational dynamics and predicts them from single structures

abstract click to expand

Protein structure tokenizers (PSTs) are workhorses in protein language modeling, function prediction, and evolutionary analysis. However, existing PSTs only capture local geometry of static structures, and miss the correlated motions and alternative conformational states revealed by protein ensembles. Here we introduce Ensembits, the first tokenizer of protein conformational ensembles. Ensembits address challenges inherent to tokenizing dynamics: deriving informative geometric descriptors across conformations, permutation-invariance encoding of variable-size ensembles, and conquering sparsity in dynamics data. Trained with a Residual VQ-VAE using a frame distillation objective on a large molecular dynamics corpus, Ensembits outperforms all related methods on RMSF prediction, and is the strongest standalone structural tokenizer on an token-conditioned ANOVA test on per-residue motion amplitude. Ensembits further matches or exceeds static tokenizers on EC, GO, binding site/affinity prediction, and zero-shot mutation-effect prediction despite using far less pretraining data. Notably, the distillation objective enables Ensembits to predict dynamics token from one single predicted structure, which alleviates dynamics data sparsity. As the field moves from static structure prediction toward ensemble generation, Ensembits offer the discrete vocabulary needed to bring dynamics into protein language modeling and design.

0

q-bio.BM 2026-05-14

Neural net predicts ER activity at 91% test accuracy

by Belaguppa Manjunath Ashwin Desai, Shreyas Murthy +4 more

Predicting Endocrine Disruptors: A Deep Learning QSAR Model for Estrogen Receptor Activity

Model trained on 224 compounds and 2944 features offers faster screening for endocrine disruptors than traditional assays.

abstract click to expand

Endocrine-disrupting chemicals (EDCs) threaten human health, ecosystems, and biodiversity by interfering with hormonal signaling pathways conserved across vertebrates. Traditional in vivo assays are costly and time-consuming, limiting their capacity to screen the growing number of chemicals. To address this, we developed a deep learning-based QSAR model to predict estrogen receptor (ER) binding molecules. Using a curated dataset of 224 compounds and 2,944 molecular descriptors and fingerprints, a deep neural network (DNN) incorporating dropout and batch normalization was trained and validated. The model achieved training and test accuracies of 96.65% and 91.30%, respectively, with an ROC-AUC of 0.81, a precision of 0.82, and a recall of 0.88 for the active class. Molecular docking against estrogen receptor (PDB ID: 5TOA) confirmed that several predicted compounds exhibited binding comparable to Estradiol, sharing key interactions. This model enables rapid screening of potential EDCs, supporting efficient chemical risk assessment and contributing to biodiversity conservation by identifying compounds that may disrupt reproduction and population stability in humans and wildlife.

0

cs.LG 2026-05-13 2 theorems

Hessian matching outperforms force matching on eight proteins

by Sanya Murdeshwar, Sanjit Shashi +4 more

Hessian Matching for Machine-Learned Coarse-Grained Molecular Dynamics

Stochastic curvature matching reduces distribution errors along slow modes by up to 85 percent without full Hessian costs.

abstract click to expand

Coarse-grained (CG) molecular dynamics enables simulations of atomic systems such as biomolecules at timescales inaccessible to all-atom (AA) methods, but existing CG neural potentials trained via force matching capture only the gradient of the free-energy surface, leaving its curvature unconstrained. We introduce a framework that augments force matching with stochastic Hessian-vector product (HVP) matching, instilling second-order curvature information into CG potentials without constructing the full Hessian. We derive a decomposition of the target CG Hessian into a model-independent projected AA Hessian, precomputed once before training, and a model-dependent covariance correction computed online at negligible cost. We construct an unbiased stochastic estimator of the Hessian-matching objective by using random probe vectors. We evaluate our method by comparing against force matching on a benchmark of nine fast-folding proteins unseen during training. HVP matching outperforms plain force matching on 8 of 9 proteins on slow-mode metrics, with reductions of up to 85% in the Kullback--Leibler divergence between the CG and reference distributions along the slowest collective mode of the largest protein. Our results demonstrate that higher-order physical supervision is a practical path to more accurate and transferable CG potentials for biomolecular simulation.

0

q-bio.BM 2026-05-13 3 theorems

Vibrational graphs predict protein functions without sequence

by Charles B Reilly

Frequency-Space Mechanics: A Sequence and Coordinate-Free Representation for Protein Function Prediction

Mode frequencies and octave couplings let a graph neural network classify GO terms on thousands of proteins using only dynamics.

abstract click to expand

Protein function prediction is dominated by representations grounded in sequence and static structure, neither of which captures the collective vibrational dynamics through which proteins act. Here we introduce frequency-space mechanics, a representational framework in which a protein is encoded as a mechanical harmonics graph (MHG): nodes are vibrational modes derived from molecular dynamics, and edges are harmonic couplings weighted by octave alignment between mode frequencies. The representation is coordinate-free, sequence-independent, scale-invariant, and inhabits a latent mechanical space in which the original atomic coordinates have been projected out. The same construction applies to any system with a tractable eigendecomposition. Trained on 5,238 SwissProt proteins under a strict 30% sequence-identity split and using no sequence information, a graph neural network over static MHGs predicts GO molecular function terms across the ontology, demonstrating that vibrational physics alone encodes broad functional class. Kuramoto entrainment of the harmonic coupling graph, formally a Hamiltonian operation over mode frequencies and directly compatible with quantum annealing hardware, improves prediction for proteins whose function depends on collective conformational dynamics. On CLIC1, a fold- and function-switching chloride channel excluded from training, entrainment amplifies channel-activity signal 7.5-fold and antioxidant signal 2.4-fold, recovering both functional states from dynamics alone.

0

q-bio.BM 2026-05-13

Alignment and optimization improve RNA-protein binding

by Roman Klypa, Alberto Bietti +1 more

Multimodal Alignment and Preference Optimization for Zero-Shot Conditional RNA Generation

Multimodal fine-tuning plus DPO on synthetic data produces RNA sequences with higher target affinity in zero-shot settings.

abstract click to expand

The design of RNA molecules that interact with specific proteins is a critical challenge in experimental and computational biology. Despite recent progress in natural language modeling and deep learning-based protein design, there remains significant room to improve the frequency of successful interactions and the authenticity of generated sequences for functional applications. In this work, we frame conditional RNA sequence generation as a multi-stage alignment problem, introducing Moirain: a suite of models optimized via multimodal supervised fine-tuning (SFT) and Direct Preference Optimization (DPO). Our approach begins with large-scale pretraining on diverse RNA corpora to capture the fundamental grammars of sequence plausibility. To achieve target-specific generation, we employ a multimodal SFT architecture that conditions RNA synthesis on protein structural and sequential features. Finally, we leverage DPO to refine the model using synthetic interaction data: taking advantage of DPO's unique ability to navigate non-aligned preference spaces, we improve functional fitness without collapsing the learned natural distribution. Extensive evaluation of the Moirain series (Moirain-Base, -Multi, and -DPO) demonstrates that our framework consistently produces novel, diverse, and biologically plausible RNA sequences with superior binding affinities compared to existing baselines.

0

cs.LG 2026-05-13 2 theorems

Lab variance sets PROTAC prediction ceiling at 0.67 AUROC

by Thor Klamt, Wolfgang Nejdl +1 more

Decomposing the Generalization Gap in PROTAC Activity Prediction: Variance Attribution and the Inter-Laboratory Ceiling

Cross-lab cascade bounds the dominant component of the novel-target gap well above binarisation effects, with few-shot retraining giving a 0

abstract click to expand

Machine-learning predictors of biochemical activity often exhibit large random-split-to-leave-one-target-out generalisation gaps that have been documented but not decomposed. We frame this as an evaluation-science question and use targeted protein degradation as the empirical test bed. PROTACs (proteolysis-targeting chimeras) are heterobifunctional small molecules that induce targeted protein degradation, with more than forty candidates currently in clinical trials; published predictors report AUROC of 0.85 to 0.91 under random-split cross-validation, while the leave-one-target-out (LOTO) protocol of Ribes et al. reduces performance to approximately 0.67. Random splits reward within-target interpolation, whereas LOTO measures the novel-target prediction that de-novo design depends on. We decompose this gap and identify inter-laboratory measurement variance as the dominant component, anchored by a within-target cross-laboratory cascade bounding the inter-laboratory contribution at 0.124 AUROC, well above the 0.05 contribution from binarisation-threshold choice. Across eight published architectures and ESM-2 protein language models up to 3B parameters, LOTO AUROC plateaus near 0.67, with a comparable plateau under SMILES-level deduplication; a 21-dimensional 2000-trial hyperparameter optimisation cannot break this ceiling, and the rank-1 single-seed configuration regresses by 0.161 AUROC under multi-seed validation, matching a closed-form selection-bias prediction (Bailey and Lopez de Prado, 2014). Few-shot k=5 stratified per-target retraining combined with ADMET features lifts 65-target LOTO AUROC from 0.668 to 0.7050, and post-hoc Platt scaling recovers raw output to within the 0.05 well-calibrated threshold. We release PROTAC-Bench (10,748 measurements, 173 targets, 65 LOTO folds), the variance-decomposition framework, the per-target calibration protocol, and the evaluation code.

0

q-bio.BM 2026-05-13

Protein embedding split yields independent structural features

by Mingqing Wang, Zhiwei Nie +3 more

Learning Protein Structure-Function Relationships through Knowledge-guided Representation Decomposition

Knowledge-guided decomposition of pretrained representations improves twelve downstream tasks, especially on novel structures.

abstract click to expand

Proteins encode diverse functions within complex three-dimensional structures, yet most deep learning representations remain highly entangled, obscuring the biophysical signals that underlie function. Here we introduce ProtDiS, a knowledge-guided framework that decomposes pretrained protein micro-environment embeddings into biologically grounded and task-relevant dimensions. Inspired by the information bottleneck principle, ProtDiS learns representations that balance informativeness and compression, yielding structural features that are more specific, independent, and information-efficient, and achieving consistent improvements across twelve downstream tasks, with the largest gains under structure-based splits. Protein- and residue-level analyses further show that ProtDiS differentiates proteins with similar folds but divergent functions and captures fine-grained biophysical signals critical. These findings suggest that knowledge-guided decomposition provides a general and interpretable approach for structuring latent spaces in protein structural modeling. The source code and implementation details are publicly available at https://github.com/AI-HPC-Research-Team/ProtDiS.

0

cs.LG 2026-05-12 Recognition

Deep learning with protein hierarchy predicts complexes better

by Ziwei Xie

Deep Learning for Protein Complex Prediction and Design

Architectures that respect structural layers plus algorithms scanning sequences for partners advance modeling for cellular function and drug

abstract click to expand

Accurately modeling and designing protein complex structures is a central problem in computational structural biology, with broad implications for understanding cellular function and developing therapeutics. This thesis investigates two fundamental aspects of this problem using deep learning: domain-specific architectures that capture the hierarchical nature of protein structures, and search algorithms that efficiently navigate the vast sequence spaces of protein complexes to identify interacting homologs for improving complex structure prediction and to design protein sequences.

0

q-bio.BM 2026-05-12 Recognition

Compact tokenizer generates plausible proteins from scratch with 10x fewer parameters

by Nabin Giri, Steven Farrell +1 more

Yeti: A compact protein structure tokenizer for reconstruction and multi-modal generation

Yeti achieves top codebook use and token diversity, letting a small multimodal model match larger systems in joint sequence and structure出力.

abstract click to expand

Multimodal models that jointly reason over protein sequences, structures, and function annotations within a unified representation hold immense potential for integrating multimodal data and generating new proteins with designed functional properties. To utilize transformer architectures, such models require a tokenizer that converts protein structure from continuous atomic coordinates into discrete representations suitable for scalable multimodal training. The quality of such models are fundamentally upper bounded by the fidelity and expressiveness of the underlying tokenized structure. However, existing tokenizers prioritize reconstruction over generative abilities. To address these gaps, we introduce Yeti, a simple and compact protein structure tokenizer based on lookup free quantization and trained end to end with a flow matching objective for multimodal learning. Compared to existing models, Yeti generally achieves the best codebook utilization and token diversity, and second best reconstruction accuracy (with 10x fewer parameters than ESM3) on diverse datasets. To validate Yeti's generative capability, we trained a compact multimodal model jointly over its structure tokens and amino acid sequence entirely from scratch, with no pretrained initialization. The resulting multimodal model generates plausible structures under unconditional cogeneration of protein sequence and structures, achieving comparable results to 10x larger models. Together, these results demonstrate that Yeti is a compact and expressive protein structure tokenizer suitable for training multimodal models that cogenerates highly plausible sequences and structures.

0

q-bio.BM 2026-05-11 2 theorems

New diffusion model generates directional agonists and antagonists

by Hanqun Cao, Aastha Pal +5 more

TD3B: Transition-Directed Discrete Diffusion for Allosteric Binder Generation

TD3B controls protein state transitions to produce binders whose functional bias is independent of binding affinity.

abstract click to expand

Protein function is often controlled by ligands that bias the direction of state transitions, such as agonists and antagonists, rather than stabilizing a single conformation. This is especially important for clinically relevant G protein-coupled receptors (GPCRs), where therapeutic efficacy depends on functional directionality. Structure-based design methods optimize binding to static conformations and cannot represent non-reversible, directional effects or systematically distinguish agonist from antagonist behavior. To address this gap, we introduce Transition-Directed Discrete Diffusion for Allosteric Binder Design (TD3B), a sequence-based generative framework that designs binders with specified agonist or antagonist behavior via a directional transition control objective. TD3B combines a target-aware Direction Oracle, a soft binding-affinity gate, and amortized fine-tuning of a pre-trained discrete diffusion model, enabling targeted agonist and antagonist generation decoupled from binding affinity and unattainable by equilibrium-based or inference-only guidance baselines. The code and checkpoints are available at https://huggingface.co/ChatterjeeLab/TD3B.

0

q-bio.BM 2026-05-11 2 theorems

BP180 homotrimer model stays folded in simulations

by Congzhou M Sha

A putative, computationally stable structure of homotrimeric BP180/collagen XVII

Molecular dynamics shows the AI-predicted structure of the skin adhesion protein remains mostly stable over 500 ns with domain-specific flex

abstract click to expand

Background: BP180, also known as collagen XVII and BPAG2 (bullous pemphigoid antigen 2), is a 180-kDa transmembrane protein within the hemidesmosomal plaque complex, and which is known to be a major antigen in bullous pemphigoid, gestational pemphigoid, cicatricial (mucous membrane) pemphigoid, and linear IgA bullous disease. Objective: At present, the 3D structure of BP180 is not known. The goal is to predict a reasonable structure for BP180 through machine learning and molecular dynamics. Methods: In this work, we use the recent Boltz-2 model to predict a putative structure for the intracellular, transmembrane, and proximal extracellular domains, including the NC16A antigenic region and a portion of its first extracellular collagenous domain, Col-15. We computationally embed BP180 in a simple phospholipid bilayer, demonstrate that the putative structure is stable using molecular dynamics, and analyze its allosteric properties. Results: The structures presented satisfy symmetry and secondary structure properties which are expected from homology modelling. Over three 500 ns trajectories, there is minor instability of the predicted globular head domain, but the homotrimer otherwise stays mostly folded. The putative NC16A domain is stiff, whereas the truncated Col-15 domain is highly flexible. There does not appear to be a nearby stable conformation distinct from the initial state. Conclusion: The structure presented is a useful starting point for targeting BP180 pharmacologically, for further experimental characterization of BP180, and for generating hypotheses regarding the relevant epitopes contributing to bullous disease. Diffusion models such as Boltz-2 and AlphaFold3 are useful, but their results must be evaluated carefully.

0

cs.LG 2026-05-11 2 theorems

Expert fusion model lifts microbial operon accuracy

by Seungik Cho

MicroFuse: Protein-to-Genome Expert Fusion for Microbial Operon Reasoning

Largest gains appear exactly where protein sequence misleads but genomic layout resolves true regulation.

abstract click to expand

Predicting microbial operon co-membership requires integrating two complementary biological signals: protein-scale molecular identity and genome-context organization. While recent biological foundation models provide powerful representations of each view independently, naive concatenation of these modalities ignores a key biological property -- protein identity and genomic context may agree when adjacent genes form a coherent functional module, or conflict when sequence similarity is misleading but genomic layout indicates independent regulation. We present MicroFuse, a protein-to-genome expert fusion framework that integrates structure-aware protein representations from ProstT5 with genome-context representations from Bacformer through a four-expert Mixture-of-Experts module (protein, genome-context, agreement, and conflict experts) with a learned soft router. Training combines binary cross-entropy with symmetric cross-modal InfoNCE alignment and disagreement-weighted supervised contrastive shaping. We further construct OG-Operon100K, a 100,000-pair scaffold-level benchmark from the OMG metagenomic corpus with biologically grounded positive and negative criteria. On OG-Operon100K, MicroFuse achieves the strongest AUROC, AUPRC, mAP, and mAR among ProstT5-only, Bacformer-only, and Concat MLP baselines. Ablations identify cross-modal contrastive alignment as the dominant component, and a hard sequence-conflict subset reveals MicroFuse's largest gains precisely in biologically ambiguous cases where protein identity alone is misleading.

0

cs.LG 2026-05-11 3 theorems

Graph pooling extracts functional protein substructures from ESM-2 features

by Siddhant Dutta, Edward Tan Beng Wai +3 more

Structural Interpretations of Protein Language Model Representations via Differentiable Graph Partitioning

SoftBlobGIN on contact maps lifts binding-site AUROC from 0.885 to 0.983 and yields active-site enriched blobs without supervision.

abstract click to expand

Protein language models such as ESM-2 learn rich residue representations that achieve strong performance on protein function prediction, but their features remain difficult to interpret as structural $\&$ evolutionary signals are encoded in dense latent spaces. We propose a plug-$\&$-play framework that projects ESM-2 representations onto protein contact graphs $\&$ applies $\textbf{SoftBlobGIN}$, a lightweight Graph Isomorphism Network with differentiable Gumbel-softmax substructure pooling, to perform structure-aware message passing $\&$ learn coarse functional substructures for downstream prediction tasks. Across enzyme classification, SoftBlobGIN achieves 92.8\% accuracy $\&$ 0.898 macro-F1. Unlike post hoc analysis of protein language models alone, our method produces directly auditable structural explanations: GNNExplainer recovers biologically meaningful active-site residues, spatially localized functional clusters, $\&$ catalytic contact patterns. On binding-site detection, SoftBlobGIN improves residue AUROC from $0.885$ using an ESM-2 linear probe to $0.983$, indicating that these structural explanations are not recoverable from language-model features alone. Learned blob partitions provide an additional layer of interpretability by automatically grouping residues into functional substructures, with blobs containing annotated active-site residues showing $1.85\times$ higher importance than other blobs ($\rho{=}0.339$, $p{=}0.009$), without any active-site supervision. Our framework requires no retraining of the language model, adds only $\sim$1.1M parameters, $\&$ generalises across ProteinShake tasks, achieving $F_{\max}$ of $0.733$ on Gene Ontology prediction $\&$ AUROC of $0.969$ on binding-site detection. We position this as an interpretable structural companion to protein language models that makes their predictions more transparent $\&$ auditable.

0

cs.CE 2026-05-11 2 theorems

SGRPO widens the utility-diversity frontier in biomolecular design

by Xinwu Ye, He Cao +6 more

Pushing Biomolecular Utility-Diversity Frontiers with Supergroup Relative Policy Optimization

The method adds set-level diversity rewards to policy optimization, improving both metrics over standard approaches in molecule and protein

abstract click to expand

Biomolecular generators are often adapted with reward feedback to improve task-specific utility, but pushing utility alone can concentrate generation on a narrow family of candidates. Maintaining diversity is difficult because sample diversity is a set-level property. We introduce Supergroup Relative Policy Optimization (SGRPO), a flexible GRPO-style framework that directly constructs rewards from set-level diversity. For each condition, SGRPO samples a supergroup of candidate sets, compares their diversity under the same condition, and redistributes the group diversity reward to individual rollouts through leave-one-out diversity contributions before combining it with rollout-level utility. This design decouples SGRPO from a particular generator, utility reward, or diversity metric, and allows instantiation with different GRPO-style approaches. We evaluate SGRPO on de novo small-molecule design, pocket-based small-molecule design, and de novo protein design, instantiating it with both GRPO and Coupled-GRPO across autoregressive and discrete diffusion generators. Across decoding sweeps, SGRPO expands the utility-diversity Pareto frontier and achieves the best frontier-level metrics relative to pretrained generators, GRPO, and memory-assisted GRPO when applicable. Our analyses further show that direct set-level diversity rewards remain effective with small groups and help preserve broader generation-distribution coverage during post-training. The code is available at https://github.com/IDEA-XL/SGRPO.

0

q-bio.QM 2026-05-11 1 theorem

R package MeTime stores all metabolomics steps in one container

by Bharadwaj Marella, Patrick Weinisch +6 more

MeTime: An R package for reproducible longitudinal metabolomics data analysis

Longitudinal studies keep data, metadata and results together so workflows stay transparent and reports generate automatically.

abstract click to expand

MeTime is an opensource R package for reproducible analysis of longitudinal metabolomics data. It builds upon a central S4 container, metime_analyser, that stores multiple datasets, associated metadata and analysis outputs, enabling unified handling of complex longitudinal studies. Analyses are constructed by piping modular functions, beginning with data transformations (mod_), followed by calculations (calc_), and optional meta-analysis (meta_), so entire workflows remain transparent and easy to modify. MeTime wraps numerous existing methods within a consistent interface, including sample and metabolite distributions, correlation and distance matrices, dimensionality reduction (PCA, UMAP, tSNE), random forest imputation and feature selection via Boruta, eigenmetabolites and WGCNA based clustering, conservation index analysis, regression models (linear, mixed effects, and generalized additive), and partial correlation networks. By retaining all intermediate results and provenance within the container, MeTime facilitates iterative exploration and ensures reproducible reporting via automatically generated HTML and PDF outputs. Comprehensive user guides, case studies and reference documentation accompany the package, making MeTime a versatile platform for longitudinal omics workflows.

0

q-bio.QM 2026-05-11 2 theorems

Models trained on real folds design RNA under mixed constraints

by Michael Lindsey

GoForth: Language Models for RNA Design under Structure, Sequence, and Coding Constraints

Encoder-decoder approach generates fast high-quality sequences from structure, sequence, and coding targets without post-filtering.

abstract click to expand

RNA inverse sequence design has broad biological and engineering applications, but computational methods for practical design queries remain limited. Such queries may impose several constraints at once, including target folds or motifs, fixed bases, and coding restrictions, while leaving arbitrary sequence and structure in unspecified regions. Because these constraints may permit many acceptable sequences, we study RNA design as a conditional generative modeling problem. The basic object is a conditional law over RNA sequences given a user-specified condition, with full inverse folding as a special case. We introduce GoForth, a forward-trained RNA language model that conditions on structure, sequence, and coding targets. The formulation separates three ingredients that are often entangled in RNA design: a sequence prior, a forward folding sampler, and a reward or likelihood oracle. We train encoder-decoder models on witnessed folds rather than on outputs from an inverse-design teacher and validate our methodology on full inverse-folding benchmarks, as well as tasks involving constraints on structure, sequence, and coding. The resulting models achieve fast and high-quality candidate generation for mixed RNA design specifications. Moreover they furnish useful semantic embeddings of design tasks and a robust learned notion of designability.

0

cs.LG 2026-05-11 Recognition

Masked-position latent prediction beats MLM on protein tasks

by Dan Ofer, Dafna Shahaf +1 more

ProteinJEPA: Latent prediction complements protein language models

Hybrid recipe wins 11 of 16 benchmarks for 150M models under matched wall-clock budgets.

abstract click to expand

Protein language models are trained primarily with masked language modeling (MLM), which predicts amino-acid identities at masked positions. We ask whether latent-space prediction can complement these token-level objectives under matched wall-clock budget. Across pretrained and random-init protein sequence encoders at 35--150M parameters, we find that the best protein-JEPA design is not all-position latent prediction but a variant: predicting latent targets only at masked positions, and retaining the MLM cross-entropy. We call this recipe masked-position MLM+JEPA. On a 16-task downstream suite (15 frozen linear probes plus SCOPe-40 zero-shot fold retrieval), under matched wall-clock budgets, this recipe wins more tasks than it loses against MLM-only continuation: 10 wins / 3 losses / 3 ties (hereafter W/L/T) on pretrained ESM2-35M, 11/2/3 on ESM2-150M while results in pretraining from scratch are mixed (6/8/2). Gains are seen for multiple models on 11 of 16 tasks, including stability, \b{eta}\beta \b{eta}-lactamase fitness, variant effect, intrinsic disorder, remote homology, enzyme classification, and SCOPe-40 fold retrieval. Tasks with more losses than wins are Fluorescence (TAPE) and Peptide-HLA Binding. All-position MLM+JEPA matches MLM-only overall but does not reproduce the masked-position gains. JEPA-only (no MLM) collapses in nearly every experiment. We conclude that JEPA, when combined with MLM, is competitive and can outperform pure MLM in pretraining and continued training, even under matched wall-clock budgets.

0

q-bio.BM 2026-05-11 2 theorems

Benchmark trains models on noisy DEL screens and tests them on real Ki affinities

by Mutian He, Hanqun Cao +5 more

CA-DEL: An Open Multi-Target, Multi-Modal Benchmark for Learning from DNA-Encoded Library Screens

CA-DEL pairs screens for three carbonic anhydrase isoforms with ChEMBL validation data to check generalization from indirect signals.

abstract click to expand

The success of machine learning in drug discovery hinges on learning the relationship between a chemical structure and its biological activity. While DNA-Encoded Library (DEL) technology can generate the massive datasets required for this task, its primary signal -- sequencing read counts -- is an indirect and often noisy proxy for true molecular binding affinity. To address the scarcity of public benchmarks for developing robust models that can overcome this data challenge, we introduce CA-DEL, a multi-dimensional public benchmark featuring screens against three homologous carbonic anhydrase isoforms. While recent benchmarks like KinDEL have introduced 3D poses for kinase targets, CA-DEL distinguishes itself by focusing on the selectivity challenge among homologous Carbonic Anhydrase isoforms (CAII, CAIX, CAXII). Unlike benchmarks relying solely on noisy enrichment scores, CA-DEL integrates a rigorous validation set of experimentally determined binding affinities ($K_i$) from ChEMBL, establishing a critical Sim-to-Real evaluation paradigm: training on noisy DEL screens and testing on high-fidelity biophysical data.

0

q-bio.BM 2026-05-08

Tree search guides diffusion to balance multiple protein design goals

by Zitai Kong, Yifan Dong +4 more

MP2D: Constrained Monte Carlo Tree-Guided Diffusion for Multi-Objective Protein Sequence Design

MP2D explores denoising paths with Pareto constraints to improve four or five conflicting properties without retraining the base model.

abstract click to expand

Designing functional protein sequences that satisfy multiple desired properties is a core research focus of protein engineering. Prior methods struggle with inability or inefficiency when dealing with numerous, often conflicting, properties. We propose Multi-Property Protein Diffusion (MP2D), a unified framework for multi-objective protein sequence optimization that integrates conditional discrete diffusion with constrained MCTS and global iterative refinement. MP2D formulates diffusion denoising as a constrained sequential decision-making process and employs MCTS to explore diverse denoising trajectories guided by Pareto-based rewards. A global iterative refinement strategy further enables repeated remasking and re-optimization of candidate sequences, while a dynamic Pareto constraint prevents candidate bloat and maintains balanced trade-offs across objectives. We evaluate MP2D on two challenging multi-objective protein design tasks: antimicrobial peptide and protein binder optimization, involving four to five conflicting properties. Experimental results demonstrate that MP2D consistently outperforms existing multi-objective baselines, achieving robust and balanced improvements across all objectives without retraining generative models. These results highlight MP2D as a practical and scalable solution for multi-objective functional protein design.

0

q-bio.BM 2026-05-07

AlphaFold guidance raises TM-scores in Phenix cryo-EM models

by Chenwei Zhang

Enhancing Cryo-EM Density Map Segmentation in Phenix for Improved Atomic Model Building

Inserting predicted structures into the segmentation step improves sequence accuracy over standard Phenix on noisy maps.

abstract click to expand

We introduce PhenixCraft, a fully automated pipeline for building atomic models from cryo-EM density maps. By integrating AlphaFold predictions, we enhance the map-segmentation step in Phenix during model building, addressing challenges posed by noise and artifacts that traditionally hinder this step. Our results demonstrate PhenixCraft's superior performance in TM-scores and sequence accuracy, significantly improving upon the limitations and inefficiencies of traditional model building using Phenix.

0

cond-mat.soft 2026-05-07

Condensin reverses DNA loops via catch-bond intermediate

by Atreya Dey, Guang Shi +2 more

Loop Extrusion Reversal by Condensin Motor is Mediated by Catch Bonds

Postulated ATP-cycle state shows lifetime that rises then falls with force, explaining observed backward steps.

abstract click to expand

Structural Maintenance Complexes (SMC) are energy consuming motors that are important in folding the genome by loop extrusion (LE) in all stages of the cell cycle. Single molecule magnetic tweezer pulling experiments have revealed that condensin, a member of the SMC family involved in mitosis, takes occasional backward steps, thus coughing up the gains in the length of the extruded loop. To reveal the mechanism of the forward and backward steps simultaneously, we developed a theory using the stochastic kinetic model and the scrunching mechanism for LE. The calculations quantitatively account for the measured force-dependent step size and dwell time distributions in both the directions. By postulating the existence of an intermediate state in the ATP-driven cycle that is poised to take a forward or a backward step, we predict that its lifetime increases as the external mechanical force increases till a critical value and subsequently decreases at higher forces. The surprising finding of lifetime increase in an active motor, at sub-piconewton forces, is the characteristic of catch bonds, known in force-induced rupture of several passive protein complexes. The identification of catch bond-like states in condensin not only expands our understanding of LE but also highlights the significance of mechanical forces in regulating genome organization.

0

q-bio.BM 2026-05-06

Boltz-2 and fine-tuned DrugFormDTA top antiviral affinity benchmarks

by Daniel C. Elton, Preston W. Estep

Benchmarking open-source tools for in silico antiviral drug discovery

Tests on 853 compounds across 16 viral targets show ML models outperform docking, with fine-tuning lifting correlation to 0.7.

abstract click to expand

Antivirals are uniquely positioned to be deployed quickly during a new outbreak, especially when repurposed from approved drugs. Yet there are no FDA-approved antivirals for the majority of viral families with pandemic potential. Here we lay out the case for investing in technologies and techniques for antiviral drug discovery and designing antiviral combinations. We present a survey of open source datasets and computational tools for in silico antiviral drug discovery, with a particular focus on the latest AI-based systems and docking tools. We then present our custom dataset of 43,005 viral protein-ligand binding measurements that we curated from BindingDB and other sources. Importantly, we found that 31% of viral protein binding data in BindingDB required polyprotein sequences to be carefully split before the data were suitable for training or testing ML models. Using our custom dataset we fine-tuned the DrugFormDTA binding affinity prediction model (Khokhlov et al. 2025). We then benchmarked 15 open-source binding affinity prediction tools on a custom test set of 853 antiviral compounds spread across 16 different protein targets from 10 virus species. Models tested include Boltz-2, GNINA, FlowDock, Interformer, AutoDock-GPU, and others. We found that Boltz-2 and DrugFormDTA ranked highest overall among ML-based approaches, and GNINA did best among docking approaches, with notable variance across specific viral proteins. Fine-tuning DrugFormDTA on our custom cleaned antiviral dataset boosted performance from $r=0.5$ to $r=0.7$. As part of this work we also compiled a library of approved drugs and a comprehensive list of investigational and approved antiviral drugs that can be viewed at https://antivirals-database.radvac.org. Together, this work provides a foundation for future work towards new tools and platforms for rapid drug repurposing and rapid design of antiviral combinations.

0