Discover, rate, and discuss biology preprints.

The paper identifies context-dependent sequence features that improve prediction of de novo binder success beyond structure and affinity.

The paper introduces LongevityBench, a benchmark to evaluate LLMs' ability to understand aging biology and predict age-related outcomes from biodata.

This paper evaluates effective connectivity metrics for reconstructing neural network topology under data limitations, finding that multivariate transfer entropy is most accurate but slow, while partial cross-correlation is efficient for larger networks.

ShapeME discovers structural DNA motifs that improve prediction of protein-DNA binding beyond traditional PWMs using ChIP-seq data.

PepHammer is a web-based tool that identifies bioactive peptides by comparing user input against databases of peptides with known bioactivities and tissue associations.

The paper benchmarks precision matrix estimation methods for differential co-expression network analysis using simulated datasets and identifies GLassoElnetFast as the most accurate method under certain conditions.

CROssBARv2 integrates heterogeneous biomedical data into a knowledge graph with embeddings and LLM-driven tools for predictive modeling and question-answering.

The paper characterizes the epigenetic features of pseudogenes across human tissues and identifies differences compared to protein-coding transcripts and lncRNAs.

The paper evaluates XAI methods on GNNs for DTI prediction, finding that consensus attributions highlight atoms in binding pockets relevant to drug-target interactions.

This study benchmarks eight LCR detection tools across the human proteome, revealing differences in detected region characteristics and offering a framework for reliable annotation.

The paper introduces Meta-Encoder, a framework that combines multiple pathological foundation models to improve downstream task performance in computational pathology.

This paper identifies 10 ultraselective Nrf2 activators from natural products using computational screening to minimize off-target effects on PXR and CYP2D6.

RapCluster provides an interactive web platform with 11 clustering algorithms to improve transparency and reproducibility in clustering analysis.

SpatialFinder uses a vision-language model with human feedback to prioritize high-value regions in H&E slides for spatial transcriptomics, improving ROI selection accuracy.

KyDab is a database of antibody discovery selection campaigns that provides full-funnel selection data for AI model development and evaluation.

U-Probe is a universal probe design platform that enables agentic, AI-assisted creation of FISH probes for spatial-omics applications, validated across multiple scenarios.

The paper investigates how sex-biased gene expression influences sex differences in gene essentiality using CRISPR screens and transcriptomic data.

This study compares and optimizes methods for analyzing cell type neighbor preferences in spatial tissue data, introducing COZI which outperforms existing approaches.

TFBindFormer improves TF-DNA binding prediction by integrating DNA and TF protein features using cross-attention, outperforming DNA-only models.

This study characterizes antibiotic resistance genes in E. coli and E. hormaechei clinical isolates from Kenya, identifying high-risk clones and plasmid-mediated resistance.