Discover, rate, and discuss biology preprints.

MetaStrainer reconstructs bacterial strain genotypes from short-read metagenomic data with higher accuracy than existing tools.

This paper introduces a network-based method to analyze sample-specific metabolic flow during ovarian follicle development, identifying key metabolites and markers for follicular maturation.

SpatialCCCbench evaluates spatial cell-cell communication methods using standardized metrics to guide tool selection and benchmark future developments.

This study benchmarks how reference genome selection affects taxonomic profiling accuracy across bacterial and viral datasets, showing context-dependent impacts on accuracy and computational efficiency.

NMFBatch reduces technical variation and preserves biological structure in proteomics data through batch correction and imputation.

This paper evaluates robust Random Forests for genomic prediction under contamination, finding data transformation as the most effective strategy for improving performance in contaminated datasets.

MarkerMatch improves CNV detection sensitivity across genotyping arrays by matching probes based on proximity, outperforming consensus probe methods.

ZipStrain enables rapid and precise strain-resolved metagenomics by analyzing 2,754 human samples to identify strain-sharing gradients and gene variation.

OncoContour provides an interactive platform for visualizing and analyzing cancer incidence data across geographic regions and demographics using a no-code interface.

This paper uses NLP to analyze 500,000 animal drug studies and links them to clinical data, finding that only 3.1% reach regulatory approval and that species diversity and blinding improve translation success.

AbSolution enables integrated analysis of sequence-derived features in immune repertoires across datasets and conditions.

The paper presents a machine learning challenge to predict gene perturbation effects on T cell differentiation, finding that feature representation from perturbational data improves performance and validating top genes experimentally.

scDIG is an R Shiny tool for interactive density-based gating of single-cell proteomic and transcriptomic data, resolving CD4+ T cell subpopulations in CITE-seq data.

The paper evaluates and ranks protein backbone generation methods using a unified benchmark, identifying top-performing models in both unconditional and conditional generation.

PATTY corrects open chromatin bias in CUT&Tag data using ATAC-seq and machine learning, improving detection of histone modifications and cell clustering in single-cell analyses.

The paper introduces N-Orbit, a mathematical model for comparing tissue microenvironments by encoding cell type composition and spatial relationships, showing improved performance over existing metrics in neighborhood discrimination and clinical prediction.

SpliceRead improves canonical and non-canonical splice site prediction using residual blocks and synthetic data augmentation, outperforming existing models across multiple metrics.

The paper introduces mmContext, a framework for multimodal contrastive learning of omics and text data, and demonstrates its utility in RNA-Seq and text integration tasks.

The paper develops a pipeline to prioritize CRISPR-Cas9 sgRNAs for HDR correction of TP53 hotspot mutations in cancer, identifying cohort-specific optimal guides.

The paper introduces atomic elementary flux modes (AEFMs) to efficiently analyze metabolic fluxes in large networks, showing they outperform traditional EFMs and reveal new insights into carbon and nitrogen metabolism in HepG2 cells.