CryoLigATE: enhancing the resolvability of cryo-EM maps in protein-ligand complexes using deep learning [new] deep learning framework enhances cryo-EM ligand densities, using diverse data improving atomic modeling in drug discovery.

MEGA-ODE: Learning Biologically Structured and Navigable Continuous Perturbation Dynamics from Sparse Omics [new] Reconstructs dynamic omics profiles, predicts unmeasured states, prioritizes virt. perturbs to navigate bio. endpoints.

Dual-Specific Antibody Design Using Artificial Intelligence [new] AI platform creates functional multibodies, simultaneously binding multiple antigens in one Fv region, for advanced antibody therapeutics.

A Blind Spot in Alignment: Quantifying Biosecurity Risks in Large Language Models [new] is addressed by a protocol evaluating LLMs' biosecurity risk generating toxin-like sequences, revealing high functional harmfulness, with a mitigation classifier.

The Cost of Binarizing Survival Outcomes in Clinical Prognostic Modeling [new] ...demonstrates it discards prognostic signal. A modified Bayesian network recovers this, urging default time-to-event methods.

A structured study of cross-condition prediction of transcriptional responses to gene perturbations [new]

Homology-Based Variant-Effect Predictors Break Down on Cytochrome P450 Pharmacogenes [new] Homology assumptions fail. Improved models lack clinical relevance; substrate-conditioned variant effects are key to accurate prediction.

Multi-modal Graph Integration for Biologically Interpretable Domain Identification Using Spatial Transcriptomics [new] integrates pathway scores & histology via affinity graphs, spatial connectivity to domainID with pathway insights.

LOCALE: Local-Alignment Embeddings for Noise-Robust DNA Search at SRA Scale [updated] ...achieves this by learning vector embeddings whose inner-product similarity ranks local alignments, trained with InfoNCE on augmented sequence crops.

Contrastive Regulatory Embeddings Attention Model for Differential Expression Prediction with Personalized Genomes: Insights and Challenges [new]

BirdCODE: Detecting bird communication at scale [new] precisely identifies and classifies over 9000 bird species' vocalizations with fine temporal detail, enabling large-scale communication studies.

Trajectory forecasting lifts the femtosecond ceiling in molecular simulation [new] enabling longer molecular dynamics simulations with accurate physical property prediction using a new deep learning method.

Explainable HGT-based framework for predicting human dark kinase protein-pathway associations by leveraging BERT-based embeddings and WGAN-GP [new]

Pervasive Backdoor Vulnerabilities in Genomic Foundation Models [new] Systematically demonstrated across diverse models & tasks via minimal poisoning with synthetic/bio triggers. Det/loc defense intro.

MHChron: diversity-balanced dataset design for robust peptide-MHC binding prediction across MHC class I and II [new] Uses sys data curation, meticulous DS bals, & rig. eval on diverse DS for broad allele & peptide length coverage.

Transfer learning from computed stability data for nanobody melting-temperature prediction [new] Enhanced by auxiliary tasks leveraging computed stability props. Eff. depends on sequence choice & the property itself, not label count.

Waltz, AlphaFold and ESMfold Predictions on a Human Prion Protein Tiled Peptide Library Highlight a C-Terminal Region with Strong Conformational Sensitivity [new]

Blending physics-based and inverse folding models to disentangle variant effects on stability and function [new] ...by decomposing variant evol costs into folding stability & functional 'dark energy' parts, revealing unique dz. sigs.

A Blind Spot in Alignment: Quantifying Biosecurity Risks in Large Language Models [new] ...quantifies LLM biosecurity risks by evaluating toxin-like sequence generation, revealing high functional harm, and introducing a specialized classifier to mitigate it.

Expanding Protein Structure Prediction into Conformational State Space [new] ...by inferring accessible states, their energetic/kinetic relationships, context, and perturbation responses, moving beyond single dominant forms.

Scaling an Autoregressive Transformer for Single-Cell Generation [new] ...studies generating single-cell gene expression vectors, characterizing biological fidelity and pretraining loss scaling, identifying a two-exponent scaling law.

Multi-scale modeling of human tissues from spatial transcriptomics with TERRA [new] ...develops a foundation model gen. multi-scale embeddings (cells, genes, neighborhoods) for zero-shot in silico pert. & predicting tissue responses.

Non-Invasive Embryo Quality Assessment via Matrix-Optimized Untargeted LC-MS Metabolomics of Spent Embryo Culture Media and Weighted Ensemble Machine Learning [new]

Spatial proteomics guided by H&E-based AI reveals recurrence-risk niches in triple-negative breast cancer [new] ...by using AI risk heatmaps as guides for spatial proteomics, uncovering localized mitotic and immune programs linked to prognosis.

Evolutionary Design of Membrane-Lytic Antimicrobial Peptides with Mixture of Experts [new] applies generative adversarial IL with diverse seq ops (insert, delete, mutate) for mechanism-driven AMP creation, enabling miniaturization.

MAXWELL: Calibrating the probabilistic outputs of protein language models to the mutation-induced stability change landscape [new] post-training method to exploit PLM probs for quantif. mutation effects on stability, enabling eng.

Interpreting Protein Language Models: high attention sites predict functional regions [new] ...categorize HA sites into structural/pathogenic groups, predict pathogenic risk and identify novel binding sites for therapeutic potential.

Scientific computing in the age of agentic AI: an exploratory field report [new] ...explores agents addressing scientific software's technical debt & engineering gaps, with 8 life science case studies on maintenance & improvement.

Kukulu: Diffusion-Based Reconstruction of Antibody CDR Loops using a Structure-Aware Joint Embedding Predictive Architecture [new] recons. CDR loops in Ab-Ag cmplxs, learning latent reps to diffuse coords, atm pres. & resid. IDs.

Directed Evolution in Codon Space [new] refines optimized therapeutic coding sequences via language model-guided synonymous exploration, modulating biologic expression without altering protein.