Discover, rate, and discuss biology preprints.

This study maps viral RNA signals across cell types and tissues in colorectal cancer, identifying significant viral associations with specific cell types and tissue conditions.

This paper introduces a family of dissimilarity measures for phylogenetic networks that better capture reticulation differences than existing methods.

DaliScope analyzes DALI structural alignments to uncover protein relationships and structural patterns, revealing unannotated domains and structurally anchored sequence signatures.

This study profiles SCN transcriptomes in soybean roots under single or combined SBA infestations, finding increased SCN gene expression in resistant cultivars under combined treatment.

CPDD adapts an autoregressive DNA language model for bidirectional phage genome modeling, achieving higher random-mask restoration accuracy and supporting fitness prediction and classification.

The paper describes an automated pipeline using a large language model to collect hepatotoxicity findings from biomedical literature, creating a large-scale knowledge base with 189,212 findings.

KLONE assigns single-cell RNA-seq cells to DNA-seq identified clones using a k-NN ensemble, achieving high accuracy and minority clone recall in colorectal cancer data.

This paper introduces HV-MultiScaleGNN, a graph neural network framework for predicting host-virus protein-protein interactions using network topology, achieving strong performance across multiple evaluation metrics.

The paper evaluates AlphaFold's predictions of transmembrane protein complexes, finding that while most show plausible membrane geometry, some have inconsistencies and limitations in predicting lipid interactions.

spSheaf infers spatial domains from raw molecular data without cell segmentation, outperforming existing methods on benchmark datasets.

The paper develops a method combining evolutionary and protein language model representations to predict protein stability changes with thermodynamic consistency, improving accuracy and physical consistency.

This paper improves SBWT-based k-mer indexes by making them faster and smaller through batching, prefetching, and orientation optimization, outperforming SSHash in speed and space efficiency.

SigDoTS is a statistical framework for spatial domain detection in spatial transcriptomics that controls for post-selection bias and improves domain recovery.

OmniT2I converts molecular profiles into images and predicts survival and missing modalities from incomplete cancer data, achieving higher accuracy than separate models.

This paper introduces a deep learning framework for identifying dynamic protein allosteric sites across different conformational states, achieving improved performance over existing methods.

The paper introduces ANA, an agentic AI framework that improves cross-site generalization in ASD fMRI classification by reducing data leakage and achieving higher accuracy than existing models.

ProClust assesses the statistical significance of clusters in scRNA-seq data to refine overclustering and distinguish biologically meaningful subpopulations.

This paper introduces PertVI, a deep generative model that disentangles treatment effects from cellular state and batch effects in single-cell perturbation data, revealing state-specific transcriptional responses.

PRISM simultaneously decontaminates ambient RNA and corrects doublets in spatial transcriptomics data, achieving high accuracy and purity in a mouse cerebellum dataset.