Discover, rate, and discuss biology preprints.

The paper presents a computational framework to identify interventions that compensate for ion channel mutations by modulating other conductances, demonstrating feasibility through simulations and a differentiable Hodgkin-Huxley model.

This study identifies concordant and discordant molecular signatures between PCOS and T2DM using direction-aware analysis of transcriptomic data from GEO datasets.

MCseg improves cell segmentation and transcript attribution in spatial transcriptomics through an AI-agent-guided workflow without requiring image-analysis expertise.

Haruka identifies condition-specific and shared spatial domains across tissues and conditions using contrastive learning, demonstrating improved detection of spatially heterogeneous perturbation responses.

This paper evaluates the effectiveness of general protein language models and negative data generation strategies in predicting TCR-peptide-MHC interactions, finding that they struggle to distinguish positive from negative TCRs.

This paper introduces a PhyloPotts model that combines Potts models with phylogenetics to improve ancestral sequence reconstruction, though it does not consistently reduce bias in phylogenetic estimates.

ClustSIGNAL improves spatial transcriptomics clustering by adapting smoothing based on neighbourhood complexity, enabling scalable analysis of large datasets.

PerturbBridge predicts single-cell perturbation responses by modeling population-level distribution shifts using a conditional latent Schrödinger Bridge framework, achieving state-of-the-art differential-expression recovery.

OmicsResonance provides a code-free web platform for scRNA-seq analysis, integrating LLM-assisted tools and interactive features to enable reproducible and transparent single-cell research.

AnewDDE integrates AI tools for biomolecular interaction modelling and drug design, demonstrating improved performance in structure prediction, binder identification, affinity estimation, and ADMET prediction.

SurfGraphPro predicts protein binding sites by integrating protein language models with geometric deep learning on coarse protein surfaces, achieving faster performance than existing methods.

The paper evaluates learned fungal ITS embeddings against correctly configured alignment and finds alignment performs better in open-world settings due to evaluation choices.

The paper uses computational methods to evaluate how SoxB thiohydrolase binds thiosulfate in soil bacteria, finding higher binding affinity and structural stability under alkaline conditions.

COCOA-Tree reanalyzes coevolution datasets to uncover evolutionary properties of coevolving residues and shows different metrics identify distinct groups.

MetaCarto automatically generates readable genome-scale metabolic maps by organizing reactions into biologically meaningful layouts, matching curated pathway representations and producing 2621 maps for 108 BiGG models.

The paper introduces an enhanced pipeline for multi-omic integration using topological data analysis to rank genes by disease relevance, validated on TCGA breast cancer data.

HR-FRGS identifies biomarkers from cancer RNA-seq data using hypergraph learning, achieving high classification performance and biological relevance.

The paper develops a machine learning framework to predict MHC class II binding for canonical and modified peptides by incorporating chemical features, improving immunogenicity risk assessment for non-canonical peptide drugs.

StriMap models TCR-peptide-HLA interactions using structural and sequence features, achieving state-of-the-art performance and identifying disease-associated peptides in ankylosing spondylitis.

EISCA and EISTA are pipelines for single-cell and spatial transcriptomics analysis that provide standardized, modular workflows for data processing and downstream analysis.