Papers for

biotech software developers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Linearknot predicts rna pseudoknots fast and accurately

LinearKnot: Linear-Time Approximation of the Pseudoknot Partition Function

Abstract: Pseudoknots, where two or more helices cross, are central to many functional RNAs, including frameshifting signals, telomerase RNA, riboswitches and viral genomes. With arbitrary pseudoknots, finding the minimum free energy structure is NP-hard, so most RNA secondary structure prediction tools only consider pseudoknot-free structures. Dirks and Pierce (DP03) pioneered the first pseudoknot partition function, which underlies NUPACK's pseudoknot model, but it takes $O(n^5)$ time and $O(n^4)$ memory, making it practical only for very short sequences. Linear-time approximate dynamic programming, as in LinearFold and LinearPartition, does not transfer easily to DP03: a partial pseudoknot leaves a gap filled by its crossing partner, whose complementary sequence may still lie ahead of the scan, so it cannot be ranked by its inside weight. We present LinearKnot, a linear-time approximation of DP03 in three passes: (a) LinearPartition gives pseudoknot-free weights; (b) a second pass builds tables that estimate each partial stem's crossing partner; and (c) a third scans a left-to-right reformulation of DP03 that reserves the 5' nucleotides in the pseudoknot helix for subsequent pairing, keeping the states with the largest estimated contribution. Without pruning, this reformulation computes DP03's pseudoknot recursion exactly, so the only approximation is which states the beam keeps. End-to-end, time grows linearly ($n^{1.01}$) and memory marginally above linear ($n^{1.15}$) from 1,000 to 9,392 nt. LinearKnot folds the 9,392 nt Rous sarcoma virus genome in 2.4 hours and 90.5 GB, where exact DP03 would need an estimated 40,000 years and 54 million GB, and matches exact DP03 base-pair probabilities on 418 PseudoBase RNAs with a median RMSD of $9.2\times10^{-7}$. Our code is available at https://github.com/gautammilan/LinearKnot.

Thu 12 NovData Structures and Algorithms
The gist
Predicting RNA shapes that include special knots called pseudoknots is very slow with current exact methods. The authors developed LinearKnot, a new way to approximate these complex shapes much faster and using less computer memory. Their method runs in nearly linear time and uses a three-step process to estimate the pseudoknot probabilities almost exactly. This allows scientists to analyze very long RNA sequences that were previously impossible to handle with exact methods.
Open → 2611.17602v1

DrugCoding automates complex drug discovery tasks with programming agents

DrugCoding: A Scientific Programming Agent for Drug Discovery

Abstract: Drug discovery involves multiple interconnected research stages, including molecular design, property prediction, and virtual screening. These processes are characterized by complex workflows, interdependent computational tasks, and diverse specialized tools, placing substantial demands on the coordination of scientific models and computational resources. However, existing large language model (LLM)-based scientific agents still face challenges in autonomous planning of complex tasks, adaptive selection of specialized models, and reliable execution of multi-stage scientific computations, limiting their effectiveness in automating complex drug discovery workflows. To address these challenges, we propose DrugCoding, a scientific programming agent system that integrates scientific task planning, adaptive model routing, multi-agent collaboration, and programmatic tool execution. DrugCoding enables autonomous task planning and dynamic model adaptation for complex research objectives. It can flexibly invoke specialized scientific models, autonomously generate and execute code according to task requirements, and leverage multi-agent collaboration to support efficient execution and reliable validation of complex scientific tasks. To evaluate DrugCoding's performance in scientific tool invocation and multi-stage task execution, we conduct multi-turn tool-calling experiments across diverse language models and scientific tool interfaces. Experimental results demonstrate that DrugCoding achieves an average session-level Pass@1 of 97.35% on the full benchmark, outperforming the strongest baseline, Codex, by 9.24%. On the challenging subset, DrugCoding achieves 97.92%, exceeding the strongest baseline.

Tue 10 NovArtificial IntelligenceMultiagent Systems
The gist
Drug discovery involves many steps and tools that can be hard to manage together. The authors created DrugCoding, a system that helps plan and handle these complex steps by writing and running code automatically. It also chooses the best specialized models for each task and uses teamwork between agents to get better results. Their tests show DrugCoding is very effective at handling multi-step drug discovery jobs, doing better than other top tools.
Open → 2611.13651v1

BridgeDock improves protein ligand docking using diffusion bridges

BridgeDock: Taming Generative Co-Folding Prior for Molecular Docking with Diffusion Bridge

Abstract: Molecular docking aims to predict the three-dimensional structure of a protein-ligand complex; in the conventional structure-based setting, it is conditioned on an input protein structure and ligand, often starting from an apo receptor in flexible docking or a holo receptor in rigid docking. Recent co-folding models exemplified by AlphaFold 3 achieve strong accuracy in protein-ligand complex prediction, but they typically generate complexes from sequence-derived inputs and therefore do not directly use apo structures as structural priors in docking, leading to a larger conformational search space and weaker robustness in difficult or few-step settings. We present BridgeDock, a framework that adapts pretrained co-folding backbones to flexible docking by modeling the apo-to-holo transition as a diffusion bridge, allowing generation from the initial protein-ligand state rather than from Gaussian noise. To make this bridge formulation compatible with pretrained denoisers, we introduce an alignment mechanism that maps bridge states to the original denoising schedule of the pretrained model. Experiments on standard flexible docking benchmarks show that BridgeDock consistently outperforms strong co-folding baselines. Moreover, BridgeDock is computationally efficient at inference time and remains effective even with very few denoising steps.

Sat 31 OctArtificial Intelligence
The gist
Molecular docking tries to predict how a drug molecule fits with a protein to help design medicines. The authors address a challenge where standard models don't use the initial protein shape well, making prediction less reliable. They created BridgeDock, which better models the protein changing shape when the drug binds by using a special mathematical tool called a diffusion bridge. This approach helps predict the protein-drug complex more accurately and quickly, even when using limited computational steps.
Open → 2611.00795v1

Single cell model fills in missing molecular data across experiments

scAnyOmics: Foundation-Model Distillation for Any-to-Any Completion of Multi-omics Virtual Cells

Abstract: Recent advances in single-cell technologies have enabled the profiling of multiple molecular modalities, including transcriptomics, proteomics, and chromatin accessibility, providing a more comprehensive understanding of cellular states. However, integrating heterogeneous modalities and predicting missing molecular information remain challenging, particularly when paired measurements are limited. Here, we introduce scAnyOmics, a generative framework that distills knowledge from a frozen single-cell foundation model into a multimodal student model. scAnyOmics integrates scRNA-seq, protein-marker measurements, and scATAC-seq into a shared representation, enabling cross-modal prediction and reconstruction of missing modalities. We assembled 690,880 cells from six publicly available studies, covering both paired and partially paired multi-omics measurements, to systematically evaluate its predictive performance. We further evaluated scAnyOmics on held-out donor data, where it outperformed baseline methods across multiple evaluation metrics. Beyond cross-modal reconstruction, we explored its applications in cell-type annotation, perturbation-response prediction, and biological age estimation. Overall, scAnyOmics provides a unified framework for integrating incomplete single-cell multi-omics data and transferring pretrained biological knowledge, supporting the development of virtual cell models across diverse cellular contexts and experimental platforms.

Fri 30 OctArtificial Intelligence
The gist
Different technologies can measure various parts of a single cell, like its RNA or proteins, but it is hard to combine these different datasets, especially when some data is missing. The authors created scAnyOmics, a computer model that learns from a large database and can predict missing molecular information for single cells by combining data from different measurement types. They tested it with nearly 700,000 cells and found it can better guess missing data than existing methods. This model can also help identify cell types, predict how cells respond to changes, and estimate biological aging.
Open → 2610.41399v1

AI agent improves drug discovery workflows across tasks

MoLRSI: A Recursive Self-Improving AI agent for Molecular Discovery

Abstract: Molecular discovery requires navigating vast combinatorial search spaces while balancing competing biological objectives. AI agents can automate expert-defined molecular design protocols, but automation alone does not address how to optimize workflows for new tasks or improve their development through experience. Here we introduce MoLRSI, an AI agent that combines systematic optimization of workflows and model architectures with persistent cross-task memory to recursively improve its search policy for molecular discovery. Workflow performance, refinement outcomes, and execution feedback guide subsequent searches, allowing the agent to improve both its workflows and its strategy for developing them. We evaluate MoLRSI in two challenging molecular discovery settings that require balancing competing properties. In antibiotic discovery, MoLRSI developed a screening workflow balancing antibacterial activity, cytotoxicity and structural novelty, achieving an experimental hit rate of 15.9%, compared with 8.7% for a previous expert-led workflow using the same training data and screening library. In de novo peptide-MHC binder design, MoLRSI developed workflows that jointly consider target binding and specificity, producing eight experimentally validated hits against MART-1 and SARS-CoV-1 pMHCs with preferential binding over tested off-targets while Claude Science completely failed. These findings demonstrate how experience-guided workflow optimization can address competing discovery objectives while improving the search strategies used in subsequent campaigns.

Thu 29 OctArtificial Intelligence
The gist
Discovering new medicines involves searching through many possible molecules to find ones that work well and are safe. The authors introduce MoLRSI, an AI agent that learns from past experiences to improve how it searches for promising molecules over time. It not only automates experiments but also optimizes its methods and models to do better in future searches. In tests on antibiotic development and peptide design, MoLRSI found more effective molecules than prior expert methods and other AI approaches.
Open → 2610.39714v1

GraphRAG simplifies interpreting gene study results with faster connected insights

Automated Enrichment Analysis with GraphRAG: Efficient Multi-Database ORA Extraction and Biological Interpretation

Abstract: Over-representation analysis (ORA) is widely used to interpret gene-expression studies, but its outputs are often fragmented across gene-set databases and presented as flat term lists that obscure gene-level links among related enrichment signals. We propose Graph-Based Over-Representation Retrieval-Augmented Generation (GBOR-RAG), an integrated and traceable framework for ORA interpretation. GBOR-RAG represents heterogeneous gene-set databases as a unified term--gene knowledge graph, where shared genes connect related terms across sources. After graph-based ORA scoring, the framework retrieves a compact connected evidence subgraph that preserves both cross-database relationships and gene-level support. The subgraph conditions an LLM to generate biological interpretations, and we introduce a claim-level faithfulness metric to assess whether generated claims are supported by explicit evidence. Experiments show that graph-based ORA preserves conventional ORA results with high consistency (mean Pearson $r=0.997$) while reducing query time by a median of $37\times$ (up to $48\times$). The retrieved subgraphs average 36.8 nodes and achieve the highest overall integration score (4.12), and GBOR-RAG attains the highest claim-level faithfulness (0.90) among the compared systems. Adding a PubMed literature layer further improves the overall interpretation quality from 4.17 to 4.75.

Thu 22 OctInformation RetrievalDatabases
The gist
Gene studies often produce complex sets of biological terms that can be hard to understand because they come from many different sources and don’t clearly show how genes connect to these terms. The authors created GraphRAG, a system that turns these scattered data into a single network linking related terms through shared genes. This network helps find important groups of genes quickly and guides an AI to explain their biological meaning while ensuring the explanations are backed by real data. Adding scientific literature references makes the explanations even better.
Open → 2610.30710v1

Magnolia framework improves computational chemistry with memory and AI

Magnolia: A Persistent-Memory Agent Framework for Computational Chemistry

Abstract: Chemistry has recently entered a high-data regime as molecular databases from high-throughput platforms or computational studies have become available. Increasing automation within the chemical sciences leads to a more digital and precision chemistry. Dedicated scientific agents are becoming central to guiding the chemist through an ever-increasing amount of data and findings, including negative ones. Here, we introduce Magnolia, a persistent-memory artificial intelligence agent framework that integrates computational protocols, from docking to quantum chemistry, lab notebooks and chat logs relying on large language models (LLMs). It relies on a three-tier memory system: session-level observations are automatically captured and distilled into structured learnings, confirmed and elevated to project-tier entries and, when human-validated, promoted to cross-project rules that actively guide subsequent work and provide direct chemical feedback. We illustrate Magnolia on four case studies: quantum-chemical evaluation of aromatic systems (COMPAS database), molecular dynamics of butane and of modified DNA, and a protein-peptide docking campaign. Across these projects, Magnolia tracked 317 runs and accumulated 116 project-tier entries, from which four cross-project rules were promoted; in controlled replays of four tasks, runs with memory passed the checks that runs without it failed (for example, ten of ten docking-configuration checks, against one to four of ten). By combining automation, reproducibility, and human-in-the-loop control, our Magnolia framework streamlines docking, molecular dynamics and quantum chemistry calculations with direct analysis and LLM-augmented feedback, offering accelerated access to computational (bio)chemistry.

Mon 19 OctArtificial Intelligence
The gist
Chemistry experiments and simulations create huge amounts of data, making it hard for researchers to keep track. The authors developed Magnolia, a tool that remembers details from these studies and learns important rules. Magnolia organizes information at different levels and uses AI to help analyze and guide new experiments, improving accuracy and repeatability. They tested it on several chemistry tasks and showed it helps avoid mistakes and speeds up research.
Open → 2610.27230v1

Multimodal model improves patient survival predictions from diverse data

HGASurv: Hierarchical Granularity-Aware Integration for Multimodal Survival Prediction

Abstract: Objective: Multimodal survival prediction jointly analyzes histopathological whole-slide images (WSIs), genomic profiles, and clinical variables to stratify patients by risk. However, existing approaches often inadequately account for granularity differences across modalities, particularly in spatial organization and feature dimensionality. To address these challenges, we propose HGASurv, which introduces a hierarchical WSI encoding module and a clinical-guided graph propagation module for granularity-aware multimodal integration. Methods: The hierarchical WSI encoding module maps WSI patch features onto a spatial grid using their original coordinates and applies parallel convolutions to construct representations at patch, local-neighborhood, and regional scales. Genomic signature embeddings then query these scale-specific WSI representations through cross-modal co-attention, producing signature-conditioned morphological features. The clinical-guided graph propagation module uses clinical variables to define patient-similarity edges and propagates fused image-genomic node embeddings with graph attention, integrating clinical information through topology rather than feature concatenation. Results: Across three TCGA cohorts, HGASurv achieved the strongest overall C-index performance, with ablation studies supporting the complementary contributions of the hierarchical WSI encoding and clinical-guided graph propagation modules. Conclusion and Significance: By introducing hierarchical WSI encoding and clinical-guided graph propagation, HGASurv unifies multi-scale pathology representation with topology-based clinical integration in a hierarchical granularity-aware framework. This design shows that matching representation and integration mechanisms to modality-specific granularity improves multimodal survival prediction.

Mon 19 OctComputer Vision and Pattern Recognition
The gist
Predicting how long patients will survive cancer can use images of tissue, genetic info, and clinical data, but these data types have different details and scales. The authors developed a system called HGASurv that carefully matches these differences by analyzing tissue images at multiple scales and combining this with genetics and clinical info using a graph-based approach. Their approach better predicts patient outcomes than earlier methods across several cancer datasets. This shows that respecting the unique characteristics of each data type helps improve survival predictions.
Open → 2610.26357v1

Selective drug design models struggle to avoid off-target effects

On the Selectivity of Generative Models in Structure-Based Drug Design

Abstract: Designing small-molecule drugs that selectively bind to a target protein while avoiding unintended interactions with other proteins is essential for reducing side effects that contribute substantially to clinical attrition. Although structure-based drug design (SBDD) has advanced rapidly with deep learning, most generative models still optimize binding to a single target and disregard selectivity due to limited training data and unknown off-target interactions. Motivated by this gap, we introduce SELECTBENCH, a comprehensive benchmark assessing the selectivity of generative SBDD models. SELECTBENCH provides three curated evaluation tasks reflecting different real-world drug discovery scenarios: literature-based case studies, safety screening panels used in pharmaceutical development, and large-scale structure-based evaluation sets. We further introduce a suite of metrics to quantify selectivity and conduct a systematic evaluation of nine representative SBDD models, including methods explicitly trained for selectivity and widely-used inference-time guidance techniques. Our results show that even selectivity-trained models achieve only modest off-target discrimination and inference-time guidance provides limited gains, underscoring the need for method improvement. SELECTBENCH provides both a characterization of where current generative SBDD stands on selectivity and an open, extensible platform to accelerate progress. The benchmark is available at https://github.com/lovetaro/selectbench.

Fri 16 OctMachine LearningArtificial Intelligence
The gist
Designing drugs that stick only to the intended proteins without bothering others is important to reduce side effects. The authors created SELECTBENCH, a new way to test how well computer programs can design such selective drugs by looking at multiple real-life situations. They tested several existing programs, including some designed to focus on selectivity, and found that these models still have a hard time avoiding unwanted interactions. This work highlights where current tools fall short and offers a shared platform to help improve future drug design programs.
Open → 2610.24124v1

Cooperative AI agents improve RNA sequence design speed and success

CocoRNA: Collective RNA Design with Cooperative Multi-agent Reinforcement Learning

Abstract: Designing RNA sequences that reliably fold into specific secondary structures is essential for understanding their biological functions but remains a challenging computational problem. We propose CocoRNA, a cooperative multi-agent reinforcement learning framework for RNA inverse design. CocoRNA simplifies the design task by decomposing it into smaller sub-problems, each solved collaboratively by multiple agents. This approach reduces the complexity of the problem and improves the exploration of design policies. During training, a centralized critic uses global structural information to guide the agents, enabling them to jointly optimize their design strategies. As a result, CocoRNA learns high-quality RNA design policies that generalize effectively to unseen structures without additional training. Experiments on the Rfam dataset demonstrate that CocoRNA substantially outperforms state-of-the-art methods in both success rate and design speed. Further experiments on other biological sequence design tasks highlight the effectiveness and broad potential of CocoRNA for complex design tasks.

Fri 16 OctMachine LearningMultiagent Systems
The gist
Designing RNA molecules that fold into specific shapes is important for biology but hard to do with computers. The authors propose CocoRNA, a system where many AI agents work together to design RNA sequences piece by piece. These agents learn from a shared guide that considers the whole RNA shape, helping them improve their strategies. Tests show CocoRNA works faster and better than other methods and can handle new RNA shapes without extra training.
Open → 2610.23820v1

DrugHarness evolves scientific skills to improve drug discovery

DrugHarness: An Evolvable Agent Harness for Computational Drug Discovery

Abstract: Computational drug discovery requires agents to compose scientific operations and adapt to changing evidence and constraints. Here we introduce DrugHarness, an evolvable agent harness coupling composable scientific execution with conditional skill induction. Its environment integrates over 80 scientific operations, biomedical resources and task-specific code, preserving associations among candidates, results and artifacts. From execution trajectories and outcome feedback, DrugHarness induces action hypotheses with observable applicability conditions, recommended actions and supporting evidence. An independent review agent evaluates candidate skills derived from these hypotheses through source verification and, when needed, targeted execution trials. Admitted skills guide subsequent tasks and are revised through feedback. Evaluation on MolBench and SMDD-Bench demonstrates capabilities in molecular screening, optimization and constrained design. Without evolution, DrugHarness exceeds reported MolClaw-CC results by 10.8 percentage points in affinity-comparison accuracy and 55% in mean top-three active hits on MolBench. Retrospective evaluation on skill-development tasks shows that evolution increases Qwen's affinity-comparison accuracy from 62.2% to 81.1% and optimization success from 79.5% to 87.2%. In a prospective HPK1 study, DrugHarness achieves a more favorable median docking score than Biomni among system-specific candidates. A synthesized compound achieves a cellular IC50 of 14.19 nM, a 2.64-fold potency improvement over the best input reference, with activity competitive with two selected compounds designed by experienced medicinal chemists. These results support conditional skill evolution for scientific agents and demonstrate a workflow connecting computational drug design with experimental validation.

Fri 16 OctArtificial Intelligence
The gist
Finding new medicines is hard because it involves many steps and adapting to new data. The authors created DrugHarness, a tool that learns and improves its ability to perform scientific tasks, like screening molecules. It tries out new ideas, checks if they work, and keeps the best ones for future use. Tests show it can find better candidate drugs than previous methods and even helped design a compound with better activity than existing ones.
Open → 2610.23791v1

Blitz speeds up protein structure prediction from 200 to 8 steps

Blitz: Accelerating the Diffusion Modules of AlphaFold3 and Boltz-2

Abstract: Diffusion-based protein structure prediction models produce all-atom structures in about 200 denoising steps, and that cost has driven a number of few-step samplers. We present Blitz, a few-step distillation framework that uses a geometric noise schedule to distribute steps more evenly across noise scales, then learns those updates through multistep consistency training. Physical steering and target-free candidate selection help reduce the remaining geometric defects at inference. We apply Blitz to AlphaFold 3 and Boltz-2, reducing denoising from 200 steps to eight. Because published few-step results use different benchmarks and scoring procedures, we compare both distilled models with their teachers and other few-step methods under a common evaluation protocol covering structural accuracy, interface quality, ligand placement, and stereochemical validity. Blitz retains teacher-level structural accuracy, with AF3-Blitz outperforming all evaluated few-step baselines and exceeding its teacher on several accuracy metrics.

Fri 16 OctMachine Learning
The gist
Predicting detailed protein shapes usually takes a long time using a process called diffusion. The authors created Blitz, a technique that speeds up this process dramatically by cutting the steps from 200 down to 8. Blitz uses a special way to space out changes evenly and trains the model to keep its accuracy over multiple steps. When tested on two advanced protein models, AlphaFold3 and Boltz-2, Blitz kept or improved their accuracy while running much faster.
Open → 2610.23703v1

ProTok creates one code system for protein features from three types

ProTok: Unified Protein Tokenization via Modular Multi-Codebook Quantization

Abstract: Protein sequence, structure, and functional annotations each capture complementary aspects of protein biology, yet existing tokenizers encode each modality independently, limiting the expressiveness of discrete protein representations. We present ProTok, a unified protein tokenizer that maps all three modalities into a modular discrete vocabulary via product quantization with per-modality codebook groups and modality dropout. A single ProTok checkpoint supports flexible inference from any modality subset without retraining. On seven StructTokenBench benchmarks spanning 14 tasks and 28 test splits, ProTok achieves an average rank of 1.4 across all structure-only tokenizers, placing first on 21 splits and among the top two on 27 of the 28. These gains come without sacrificing reconstruction: ProTok's tokens reach 0.939 lDDT on CATH 4.2, showing that multi-codebook quantization and multi-modal training yield tokens that are both reconstructable and informative. ProTok's structure tokens further lead all baseline structure tokenizers on zero-shot functional retrieval across CATH, Enzyme Commission, and Gene Ontology label hierarchies. The vocabulary also supports generation: a compact language model trained from scratch over ProTok tokens jointly generates backbone and sequence, with 52% of sampled backbones designable versus 21% for a parameter-matched GeoBPE baseline. ProTok shows that structure, sequence, and function can share one discrete vocabulary, motivating protein language models that reason over a unified token stream.

Tue 13 OctMachine Learning
The gist
Proteins can be understood through their sequences, shapes, and functions, but existing methods treat these separately. The authors propose ProTok, a way to combine all three into one set of codes (tokens) that computers can use flexibly. This helps machines better learn and generate proteins by understanding multiple aspects at once. Their tests show ProTok outperforms other methods in representing protein structures and predicting protein functions without losing detail.
Open → 2610.18481v1

RNA sequence and structure designed together with new generative model

La-Ribo: RNA Co-Design via Geometry-Latent Flow Matching

Abstract: RNA function arises from the coupling of nucleotide sequence and three-dimensional structure, motivating their joint design. Coordinating global folding with nucleotide-level detail remains challenging under limited structural supervision. We introduce La-Ribo, a generative framework for RNA sequence-structure co-design via geometry-latent flow matching. La-Ribo retains a sparse phosphate-sugar--base scaffold and encodes nucleotide identity and local conformation in residue-wise latents. A shared flow network generates both jointly, and an RNA-specific decoder then reconstructs all heavy atoms. To expand supervision, we construct a quality-controlled corpus of 168,561 RNA structures, integrating experimental data with predictions from three folding models, including 10,631 MSA-supported structures generated in this work. La-Ribo improves designability and codesignability over the evaluated baselines across sampling budgets and two refolding models, and the same prior supports scaffold-conditioned inverse folding without additional training.

Thu 8 OctArtificial Intelligence
The gist
RNA molecules work by their sequence and shape cooperating closely, but designing both at once is tough. The authors developed La-Ribo, a tool that generates RNA sequences and their 3D structures simultaneously using a special network that understands geometry. They trained it on a large dataset combining experimental and predicted RNA structures, improving the quality of designed RNA molecules compared to older methods. This tool also works for designing sequences based on a fixed structural scaffold, making RNA design more flexible.
Open → 2610.12236v1

CryoCue improves protein structure mapping using hetero density data

Learning from Hetero Density for Cryo-EM Protein Reconstruction

Abstract: Reconstructing protein structures from cryo-electron microscopy (cryo-EM) maps is essential for understanding macromolecular assemblies. Although learning-based methods have improved protein reconstruction, information from hetero components remains underused. Our analysis finds both false predictions and reference protein sites near hetero components; filtering nearby candidates can improve or impair chain construction. We introduce CryoCue, a framework that uses hetero information to guide protein reconstruction. An anchor-supervised detector learns hetero representations across five component classes. Multiscale hetero features guide backbone localization, while predicted hetero candidates condition structure refinement through their class, confidence, and frame-relative geometry. Experiments show that CryoCue improves backbone localization near hetero components and achieves more accurate protein structure reconstruction.

Thu 8 OctMachine Learning
The gist
Proteins are essential building blocks in our bodies, and understanding their 3D shapes helps scientists learn how they work. Cryo-electron microscopy (cryo-EM) gives images of these proteins, but it's tricky to piece them together correctly, especially near other molecules called hetero components. The authors created CryoCue, a method that learns to recognize these nearby molecules and uses that information to build more accurate protein shapes. Their tests show that CryoCue improves how well protein backbones are located, making the overall structure reconstruction better.
Open → 2610.11403v1

Radioactive watermark protects protein design models from misuse

MARCO: The Radioactive Watermark for Protein Generative Models

Abstract: Protein Generative Models (PGMs) have revolutionized structural biology by enabling the design of complex 3D protein structures from sequence data. However, this breakthrough introduces a dual-use challenge, exposing high-value PGMs to economic risks like unauthorized model extraction and biosecurity threats such as biohazard synthesis. To mitigate these threats, we propose \textbf{MARCO} (\textsc{COnformation waterMARk}), the first radioactive watermarking framework specifically tailored for PGMs. MARCO establishes a Dual-Layer defense that simultaneously protects intellectual property and ensures the forensic traceability of potential biosecurity misuses. (i) To preserve efficiency, MARCO iteratively embeds watermarks during diffusion reverse denoising via an auxiliary encoder-decoder, allowing the original PGM parameters to remain frozen for broad compatibility. (ii) To preserve biophysical fidelity and maximize robustness, we employ specialized loss functions targeting $C_α$-atom pairwise distances and torsion angles ($ψ, φ$) within an adversarial training framework integrated with stochastic attack simulations. (iii) Crucially, MARCO exhibits ``radioactivity'' where the watermark automatically transfers to the outputs of any pirate models trained on the watermarked data, effectively countering model extraction attacks. Comprehensive experiments demonstrate that MARCO achieves superior fidelity and robustness while successfully validating watermark transferability.

Tue 6 OctCryptography and SecurityArtificial Intelligence
The gist
Designing new proteins with AI models can create risks if the models are stolen or misused, such as making dangerous substances. The authors developed a new technique called MARCO that hides a special, invisible mark inside the generated proteins to prove where they come from. This mark stays even if someone tries to copy the model or tamper with the data. It helps protect the creators' work and makes it easier to track if the technology is used for harmful purposes.
Open → 2610.08316v1

Controlling false acceptance in adaptive generator verifier loops

Valid Stopping in Adaptive Generator-Verifier Loops

Abstract: Numerous agentic workflows are based on a generator-verifier loop: a generator proposes candidates, a cheap verifier scores them, and the workflow terminates when a proposal is verified as good enough. The verifier typically proxies a more costly ground-truth oracle, and as the generator searches adaptively against it, false acceptances may accumulate. Proposals can pass the proxy but fail under the costlier ground-truth check. We study when to stop these loops while controlling the false discovery rate of the accepted proposals. Our construction introduces tools of independent interest in distribution-free statistical testing and conformal risk control, including analysis of $e$-values constructed through index betting and a novel conformal risk control procedure for non-monotone losses. We validate the approach in synthetic settings and on a protein-design benchmark.

Mon 5 OctArtificial IntelligenceMachine Learning
The gist
Some computer processes repeatedly propose solutions and then check if they are good enough to stop. However, the checking step is usually a fast guess that can make mistakes, letting bad solutions slip through. The authors studied how to know when it's safe to stop these loops without accepting too many bad solutions. They created new math tools to make sure the accepted solutions are reliable and tested their ideas on simulated data and protein design tasks.
Open → 2610.06432v1

Biological reasoning models often ignore their biological inputs in tasks

When Do Biological Reasoning Models Use Their Biological Inputs?

Abstract: Biological reasoning models use post-training to connect LLMs to biological foundation model representations and biological text. Their benchmark accuracy is taken as evidence that LLMs reason over these inputs. We test this assumption in six biological reasoning models across DNA, protein, and single-cell tasks. We perturb one biological input while holding the query and other inputs fixed, construct evidence conflicts that pair the foundation model representation of one genome, protein, or cell with the text of another, fit linear probes to the representations the language model receives, and analyze reasoning traces against the biological inputs. Evo2 and ESM3 contribute little to BioReason and BioReason-Pro performance on the evaluated tasks. Shuffling the DNA sequence barely changes BioReason disease prediction accuracy, and in evidence conflicts the two models follow the text in 97.9% and 99.7% of cases. Linear probes trained on the Evo2 and ESM3 representations predict the task targets, so these foundation models encode information relevant to the task, but provide limited overall performance improvement to BioReason and BioReason-Pro. In contrast, foundation model inputs contribute to ChatNT, Prot2Text-V2, and CellWhisperer performance, and differentially expressed genes in the gene sentence contribute to Cell2Sentence-Scale performance. Across SFT and RL checkpoints of BioReason-Pro and 42 BioReason checkpoints, increases in accuracy do not imply greater performance contributions from biological inputs. BioReason traces misstate nucleotide changes, while BioReason-Pro traces describe functions omitted from final predictions under evidence conflicts. We find that current post-training strategies do not ensure that foundation model representations contribute to task performance.

Thu 1 OctMachine Learning
The gist
Some computer models try to solve biology problems by combining information from large language models and biology-specific data. This study tested six such models on tasks about DNA, proteins, and cells to see if they really use their biological information. The researchers found that some models mostly rely on language text rather than the biology data, even when that biology data contains useful info. Other models do use biological inputs more to perform better. Overall, just adding biological data to these models doesn’t guarantee they use it to improve their answers.
Open → 2610.00898v1

Protein fitness prediction improves by combining models with evolutionary data

A General Harness for Protein Foundation Model Fitness Prediction

Abstract: Accurate fitness prediction is central to protein engineering and understanding sequence-function relationships. With advances in deep learning, protein foundation models (PFMs) have become widely used for this task. Recent analyses, however, show that these models share preferences reflecting their training corpora, while unreliable inputs can further distort fitness predictions. Family-specific evolutionary evidence and structural context can help address these limitations by providing complementary constraints on model scores, motivating VenusREM-Harness (VRH), a general, model-agnostic, training-free Retrieval-Enhanced Mutation harness. It fuses frozen model scores with multiple sequence alignment (MSA) evidence according to model uncertainty, then applies gated background correction and score shrinkage based on structural confidence and solvent exposure. Across 1,211 assays and 3.1 million measured variants from ProteinGym, VenusMutHub, and the newly curated viral benchmark VenusViroHub, all 71 configurations improve Spearman correlation on all 3 benchmarks by 0.073 on average, with broad gains across 5 metrics. Extended analyses relate retrieval gains to model-MSA preference differences, assess domain-level gains and immune-escape cases, and quantify computational speedups. Built with VRH, VenusREM2 is the first to rank highest in all function, taxon, MSA-depth, and mutation-depth categories, with a ProteinGym Average Spearman of 0.556, 0.038 above the prior best.

Mon 28 SeptArtificial Intelligence
The gist
Predicting how changes in proteins affect their function is important for designing new proteins. The paper describes a method that improves predictions by combining existing protein models with information about how proteins have evolved and their 3D structure. This approach helps correct biases and uncertainties in the models and leads to better accuracy across many tests. The authors built a tool called VenusREM2 that outperforms previous methods in predicting protein fitness.
Open → 2609.34654v1

Coding sequence optimization runs faster and uses less memory

SparseDesign: Scaling Exact Coding-Sequence Design

Abstract: Exact optimization of synonymous coding sequences under a joint folding-energy and codon-usage objective is limited by expensive dynamic-programming splits and large working sets. \textsc{SparseDesign} applies candidate sparsification to the multiloop recurrence of a Turner~2004 dangle-0 solver over a weighted codon automaton. A direct branch is retained only when it strictly improves on every partitionable or endpoint-unpaired realization of the same endpoint states. We prove equivalence to the dense recurrence in real arithmetic, under an explicit scalar branch-interface assumption. With $N$ automaton states, edge set $E$ and $Z$ retained candidates, multiloop work is $O(N^2+N|E|+NZ)$; worst-case time remains cubic for bounded-width automata and total memory remains quadratic. Endpoint ownership permits parallel candidate construction without locks. While synthetic stress families can benefit little from sparsification and exhibit near-quadratic candidate growth, natural proteins show substantial candidate-count reductions. In our 7,600-task campaign, the 2,000-protein human-table panel has median retention of only 3.53\% at $λ=0$ and 2.15\% at $λ=4$, corresponding to approximately 28.3-fold and 46.4-fold reductions relative to all feasible direct intervals. The primary performance experiments use an AMD EPYC 7313 server. For human Dp427c (11,031 nt, $λ=0$), 16-thread packed \textsc{SparseDesign} achieves five-run medians of 236.54 seconds wall-clock time and 14.43 GiB peak RSS. Compared with the single-thread local dense LinearDesign fork on the same server (4,912 seconds, 402.10 GiB RSS), this gives a 20.8-fold wall-clock speedup and a 27.9-fold peak-memory reduction. On a Core i9-14900KF commodity PC with 64 GiB RAM, the same input, layout and thread count achieve 126.42 seconds and 14.43 GiB RSS.

Sat 26 SeptData Structures and Algorithms
The gist
Designing DNA sequences that code for the same proteins but fold into specific shapes is very hard and uses lots of computer time and memory. The authors introduce SparseDesign, a new way to cut down the number of possibilities the program checks without missing better solutions. This method speeds up the design process a lot and cuts memory use by more than 20 times on real protein examples. SparseDesign works well in practice, especially for natural proteins, enabling faster and more efficient genetic sequence design.
Open → 2609.32308v1

Graph neural network predicts protein membrane structure from atoms

Predicting Transmembrane Protein Topology from 3D Structure

Abstract: This paper presents a novel approach to infer protein topology using the state-of-the-art graph neural network (GNN), SchNet. The model is trained on the same dataset used to develop the recent DeepTMHMM model with 5-fold cross-validation. Unlike the conventional approaches based on using only the protein sequences or the $α$-carbons as features, we have decoded our classifier in this way, so all atom-level embeddings are used. Without applying any pre-trained weight, the final results have shown great potential that GNNs can be used for topological predictions.

Thu 24 SeptArtificial Intelligence
The gist
Figuring out how proteins fit into cell membranes is important for biology and medicine. This paper shows a new way to predict these protein shapes using a graph-based AI model called SchNet, which looks at all the atoms in the protein. The authors trained their model on data previously used for another method called DeepTMHMM and got promising results. Their approach is different because it uses the full 3D structure instead of just protein sequences or a few atoms. This suggests that graph neural networks can be helpful in understanding protein membrane topologies.
Open → 2609.30446v1

MorphoOrgaAgent automates organoid analysis with natural language input

MorphoOrgaAgent: A Foundation-Model-Based Multi-Agent System for Autonomous Organoid Analysis

Abstract: Organoids are three-dimensional tissue models whose morphology provides important insights into tumor development, disease progression, and drug testing. Extracting these morphological features relies heavily on manual segmentation, which is time-consuming and labor-intensive. Furthermore, performing quantitative statistical analysis typically requires custom coding skills and a mathematical background, presenting a major barrier for experimental biologists. To address these challenges, we introduce MorphoOrgaAgent, a multi-agent framework that achieves zero-shot organoid segmentation, automated data analysis, and report generation based on natural language input. The framework consists mainly of three core components: a TaskUnderstandingAgent that identifies requested measurements and visualization types; a hybrid segmentation module that combines Cellpose-derived geometric prompts with text prompts to guide SAM3 for zero-shot organoid instance segmentation; and a ReportAgent that computes quantitative metrics and compiles them alongside generated visualizations into a structured report. We further introduce MorphoOrgaVQA, a benchmark designed for quantitative evaluation of agent systems in organoid morphology analysis. Experimental results demonstrate that MorphoOrgaAgent handles both explicit and descriptive user requests, produces measurements closely matching ground truth, and generates complete analysis reports without requiring manual programming. The complete source code and MorphoOrgaVQA benchmark are publicly available at https://github.com/peng-lab/MorphoOrgaAgent.

Tue 8 SeptMultiagent SystemsComputer Vision and Pattern Recognition
The gist
Studying tiny 3D cell clusters called organoids is important for understanding diseases and testing drugs, but analyzing their shapes usually takes a lot of manual work and coding skills. The authors created MorphoOrgaAgent, a computer system that understands what measurements and images a user wants from simple language, then automatically segments organoids from microscope pictures and generates detailed reports. This system works without needing prior training (zero-shot) and matches expert measurements closely. The authors also provide a benchmark to test such systems, and they share their code for others to use.
Open → 2609.08696v1