Discover, rate, and discuss biology papers.

This paper identifies context geometry compression in virtual cell models and shows that recoverable biological resolution, not measurement dimensionality, determines the accuracy of perturbation-response predictions.

LOL identifies phenotype-associated patterns and detects confounding factors in GEO transcriptomics data through leakage-safe preprocessing and diagnostic checks.

The paper introduces an R package called sandwrm to estimate neighborhood size and species-level genetic diversity using a Bayesian approach based on the Wright-Malecot model.

FAMUS improves protein annotation by using contrastive learning to compare query sequences against multiple profiles, outperforming existing tools in KEGG and PANTHER annotations.

The paper shows that applying a metacell model to scRNA-seq data results in PCA embeddings that follow a Gaussian mixture model, but underestimates variance in real datasets due to unaccounted gene correlations.

NYX provides fast, lossless compression for common omics data formats, achieving better compression ratios and speed than existing codecs.

The paper evaluates the provenance-to-function gap in antidiabetic peptide prediction by re-evaluating benchmarks with homologous peptide grouping and introduces the ADP-Hybrid baseline classifier.

SFUMATO provides a Bayesian probabilistic framework for spatial transcriptomics that captures uncertainty and enables continuous visualization of transcriptional domains.

DensiTree 3 is an interactive tool for comparing pairs of tree sets, providing visualizations and CCDs to assess differences in tree distributions and clade support.

The paper identifies miRNAs in tracheal and salivary samples of preterm infants that predict BPD-related pulmonary hypertension with improved accuracy when combined with clinical variables.

QuImputer formulates genotype imputation as a QUBO/Ising problem using LD and allele frequencies, evaluated on rice data with classical and quantum approaches.

MTB-KB consolidates fragmented MTB research into a structured knowledgebase with a knowledge graph, enabling exploration of MTB-host interactions and treatment strategies.

SSAlign improves protein structure search speed and sensitivity by using language models and a two-stage alignment strategy, outperforming Foldseek on large datasets.

RAMP improves cross-platform proteomics analysis by preserving protein expression order, enhancing cell-type annotation and prediction across multiple cancer types.

The paper compares gene expression in healthy and diseased tropical house crickets infected with Cricket iridovirus (CrIV) to identify molecular differences associated with disease symptoms.

The paper predicts quaternary structures of viral proteins using AlphaFold2 and AlphaFold-Multimer, identifying new insights into vaccinology and protease targets.

This paper introduces LARK, a framework that learns chemical representations from reactions, showing improved performance in reaction prediction and molecular property prediction.

SCITRAM predicts master transcription factors from single-cell transcriptomics data by reconstructing gene regulatory networks and modeling transcriptional cascades.

This paper introduces a quantum Hamiltonian-based model for gene regulatory network inference, showing improved performance on synthetic and real single-cell RNA-seq data.

This paper shows that adding biological evidence to large language models improves zero-shot protein variant prediction without retraining the model.