Discover, rate, and discuss biology preprints.

AnnotateMissense integrates and benchmarks missense variant pathogenicity predictors using genome-wide data, achieving high performance with XGBoost.

The paper introduces SNI, a framework to identify atypical protein assemblies, and applies it to NLRs, finding an undecameric assembly in NRC7 orthologs.

This paper trains species-specific transformer models for bacterial genomic surveillance, showing they can cluster genomes, identify emerging lineages, and predict antibiotic resistance gene uptake.

spatiAlytica enables non-programmer biologists to perform interactive spatial omics analysis using natural language and a viewer-centric agentic system.

The paper identifies novel biomarkers for bladder outlet obstruction using transcriptomic and proteomic data from rodent models and human patients.

This study evaluates gene embeddings from large language models and finds that functional descriptions yield high-quality, biologically meaningful representations.

The paper introduces an automated AI pipeline for correlating chemical and ultrastructural images to enable organelle-specific molecular imaging in cells and tissues.

FastDedup is a fast and memory-efficient tool for read deduplication that outperforms existing tools on both paired-end and single-end data.

This study evaluates CBMAP for scRNA-seq visualization, showing that clustering strategies significantly affect embedding quality and that Secuer improves global structure preservation.

MeiCOfi detects meiotic crossovers in diverse genomes with high resolution and low coverage, showing reliability across species and ploidy levels.

DoFormer improves perturbation prediction by using a causal Transformer that distinguishes observational from interventional regimes without DAG assumptions.

MOSAIC identifies driver mutations in cancer by accounting for patient subgroup heterogeneity in mutual exclusivity analysis, improving detection of subgroup-specific patterns.

This paper introduces CompBioBench, a benchmark for evaluating agentic systems in computational biology, showing strong performance across diverse tasks.

Ankh-score improves sequence alignment over AlphaFold3 structural alignments and traditional BLOSUM methods on benchmark datasets.

This paper introduces repare, a software method to automatically reconstruct pedigrees from ancient genomes using pairwise kinship relations, showing improved flexibility and accuracy over manual methods.

The paper introduces a large annotated cryo-EM particle dataset and validates it through reconstruction and model training.

The paper introduces a statistical model to test functional effects in spatial transcriptomics data by analyzing gene expression patterns across space.

EvoRMD predicts RNA modification types using a model that integrates sequence data and biological context, achieving strong performance and interpretability.

NeuroCDS improves CDS annotation by integrating local and global neural network representations with structural constraints, achieving high accuracy on full-length transcripts and complex datasets.

TorchLIMIX accelerates multivariate GWAS using GPU computing, finding 37 additional SNPs in Arabidopsis metabolic traits compared to univariate GWAS.