Discover, rate, and discuss biology preprints.

SpaAxFormer improves spatial transcriptomics by integrating gene-level and cell-level representations, achieving better imputation and spatial-domain identification in some datasets.

This study introduces RSMBind, a deep learning model for predicting RNA-small molecule binding sites, achieving high accuracy on benchmark datasets and demonstrating the utility of RNA language models in the absence of structural data.

This paper introduces PULSE, a context-aware steering method for protein design that improves optimization outcomes by considering sequence-specific directions, and introduces LDS to evaluate trajectory reliability.

SubCellSeg is a 3D segmentation model that outperforms structure-specific models by jointly training on diverse subcellular structures and achieving zero-shot generalization.

CellVocal links spatial context to transcriptional dynamics by integrating spatial RNA velocity with cell-cell communication to attribute cell-state transitions to signalling modules and neighbouring cells.

This study identifies multiple Drosophila serpins with characteristics similar to human SERPINE1 using a multi-omics approach and comparative analysis.

SIFT selects reference haplotypes for genome patching using compressed indexing and pseudo-MEM queries to improve T2T imputation efficiency.

The paper extends a framework to generate atomic resolution ensembles of intrinsically disordered proteins using six AI structure predictors and compares their performance against experimental data and simulations.

This study evaluates genomic language models' ability to generate taxonomically informative embeddings of viral fragments, finding that some models outperform traditional methods on reference and novel datasets.

The paper selects diverse protein targets for ligand binding data generation by prioritizing pocket similarity over sequence or fold similarity, increasing coverage of the human pocketome.

CellChat Hotspot improves the interpretability of cell-cell communication networks by integrating gene set enrichment and context-driven filtering, identifying recurrence-associated patterns in colon cancer.

This paper describes the design of an IgE disruptor antibody using Claude to remove IgE from FcεRI, showing improved activity compared to omalizumab in vitro.

DE-Rigor-Agent automates differential expression analysis for bulk RNA-seq by evaluating method suitability, reducing false discovery rates and improving statistical power across diverse datasets.

This study investigates the role of nuclear-encoded mitochondrial genes in ALS, finding enrichment in mitochondrial translation pathways and altered co-expression patterns, but no clear link between genetic risk and tissue expression changes.

RawMod detects epigenetic DNA modifications without requiring control data or prior knowledge of modification types, achieving high accuracy across multiple modification types.

This paper develops efficient algorithms for detecting long exact matches between genome assemblies in compressed space and sublinear time, revealing novel structural variations in centromeres.

LingoCell interprets gene perturbations using natural language and tissue niches, revealing synergistic disease responses and shared fibroblast programs in cancers.

DecoyOpt improves EC number prediction by distinguishing active enzymes from decoys, achieving high accuracy and low false-positive rates.