Discover, rate, and discuss biology preprints.

GraphPop reduces population genomics complexity by using graph databases, enabling faster analysis of large datasets and revealing population structure in rice and humans.

HARVEST extracts structured bioactivity data from pharmaceutical patents using AI, producing millions of records and revealing gaps in existing databases.

Hybrid Gated Fusion improves protein function annotation by integrating sequence, structure, text, and interaction data using bilinear gating and auxiliary supervision, achieving state-of-the-art results on CAFA3.

The paper benchmarks transcript classifiers to identify ambiguous mRNA-lncRNA classifications by analyzing uncertainty and feature contributions.

GraphMana manages population genomics data using a graph database, enabling efficient incremental updates and export to multiple formats.

This paper introduces SMILE, a diffusion model for converting H&E stains to multiplex IHC images, showing superior performance in pancreatic islet analysis and breast cancer stain conversion.

This paper introduces UDAL, an active learning strategy to reduce wet-lab costs for TCR-pMHC binding validation by efficiently selecting candidate pairs for experimental testing.

The paper presents a computational workflow for designing nanobody binders against a novel cancer target, achieving sub-nanomolar affinities without experimental structure or prior antibody data.

NetMHCIIphosPan predicts HLA class II antigen presentation of phosphorylated peptides using immunopeptidomics data, outperforming existing tools.

RRE improves de novo repeat libraries by recursively extending repeat models using profile HMMs, resulting in longer and more comprehensive repeat models.

This paper benchmarks RNA modification detection tools on bacterial rRNA using Oxford Nanopore DRS, finding that no single tool recovers all sites and that combining tools improves performance.

The paper introduces a compression and k-mer based method for anti-cancer peptide classification, achieving better performance than deep learning models without requiring custom features or pre-trained models.

GANGE is a deep-learning system that generates accurate genomic sequences from low-coverage, error-prone reads, reducing sequencing costs and enabling regulomics research without genome data.

FairTCR improves TCR-pMHC binding prediction equity across HLA alleles and cohort strata by reducing disparity in average-worst-group AUPRC.

The paper introduces a geometric framework to represent multi-omic signatures as multidimensional objects, revealing discordance and structural patterns in cancer metabolic circuitries.

PathwaySeeker uses AI to reason about organism-specific metabolic networks using proteomic and metabolomic data, distinguishing confirmed reactions from hypotheses.

ABB4-STEROIDS generates antibody conformational ensembles with state-of-the-art accuracy, outperforming existing methods in experimental benchmarks.

This study compares neural network-based and other machine learning models for automated function prediction in yeast using co-expression data and updated Gene Ontology annotations, finding neural networks more accurate and robust.

This study evaluates 11 genomic language models and finds they offer limited advantages over a random baseline in regulatory decoding tasks.

ProteomeScan identifies protein targets for drug-like molecules through proteome-wide docking and analysis, showing improved accuracy over random baselines.