Discover, rate, and discuss biology preprints.

MetaboCensoR filters untargeted LC-MS metabolomics data to reduce redundancy and improve downstream analysis while retaining known analytes.

GO enrichment of proteome-scale docking profiles shrinks the candidate target list: a held-out GO gate kept 11.32% of known targets in 1.26% of candidates; additive reranking was mixed.

This paper develops a graph-based QSAR pipeline to predict in vitro Caspase-3/7 activation and in vivo hepatotoxicity, showing improved performance over classical models.

The paper identifies reduced SST mRNA and depletion of specific upper-layer Sst interneurons in schizophrenia using single-nucleus RNA-seq and spatial transcriptomics.

StackHPpred identifies peptide hormones from amino acid sequences using a stacking-based ensemble framework, achieving high accuracy on independent datasets.

Songbird is a single cell whole genome sequencing copy number caller that detects whole genome duplication and outperforms existing tools in breakpoints identification and true copy number detection.

MINT infers cell-resolved spatial transcriptomes from paired histology and Slide-tags data, improving alignment and expression recovery in lung adenocarcinoma.

The paper introduces a Bayesian framework to infer latent infection risk patterns from virus neutralisation assay titration data, revealing higher and more persistent antibody concentrations for EV-D68 compared to other viruses.

The paper provides a uniformly processed compendium of alternative splicing profiles from pediatric cancer and non-diseased tissues, enabling large-scale analysis of splicing aberrations in pediatric cancers.

ANIMA predicts protein-protein interactions across species using an AI model trained on diverse animal species, achieving high accuracy in under-researched species.

This study uses machine learning to distinguish rare disease cases from controls using observed and excluded phenotypes, finding that Random Forest performs best and both specificity and occurrence count are important for feature selection.

This paper introduces a counterfactual explanation algorithm for graph neural networks to explain patient outcome predictions in patient similarity networks.

The paper identifies a transcriptional continuum from normal to lesional skin in psoriasis using single-cell RNA sequencing and trajectory analysis, revealing cell-type-specific state transitions.

StressNET infers intercellular mechanical stresses from tissue geometry using a Graph Neural Network, achieving state-of-the-art performance in zebrafish and Xenopus embryos.

Gravlax archives single-cell RNA-seq data in a compact format that preserves molecular evidence, enabling reanalysis under new annotations and discovery of unannotated features.

The paper presents a dynamic pocketome analysis of TAAR receptors using simulations, revealing receptor-specific ligand-receptor interplay and plasticity in binding sites.

The paper curates a single-cell RNA-seq catalog with known coregulation relationships to evaluate normalization methods for identifying gene coregulation.

CDSeq2 improves expression deconvolution and eQTL analysis by integrating Bayesian deconvolution with prioritization of recoverable profiles in bulk RNA-seq data.

SVPG improves structural variant detection using pangenome references and accelerates pangenome graph construction from long-read data.

stEDGE reconstructs hierarchical spatial domains and transition interfaces in spatial transcriptomics by using edge-guided methods to capture multiscale organization beyond flat partitions.