Discover, rate, and discuss biology preprints.

A Local Interaction Score (LIS), computed from AlphaFold-Multimer predicted aligned error, separates curated fly and human protein interactions from controls better than ipTM, recovering small or flexible interfaces.

12.4 million single-nucleus transcriptomes from 83 staged mouse embryos (E8 to P0) yield 190 cell types, a zygote-to-pup cell-type tree, and rapid postnatal transcriptional shifts.

Chromatin regulators including EZH2, EHMT1/2, and DOT1L establish an epigenetic barrier in human neural progenitors that poises maturation genes and sets the protracted pace of cortical neuron maturation.

Chromosome-scale inshore hagfish genome places 1R before the cyclostome-gnathostome split and 2R in the gnathostome stem, and reveals two cyclostome-specific duplications (CR1, CR2).

MERFISH profiling of >1,100 genes in ~8 million cells spanning the whole mouse brain, integrated with scRNA-seq, maps >5,000 clusters into the Allen Common Coordinate Framework.

ProteinGym unifies over 250 deep mutational scans and ~65k clinical variants to benchmark over 70 models in zero-shot and supervised settings; TranceptEVE ranks best overall.

Rare-variant depletion in 1kb windows of 76,156 gnomAD genomes, versus feature-adjusted mutation expectations, marks regulatory elements and outperforms seven alternative scores at flagging functional non-coding variants.

SigProfilerAssignment refits known substitution, doublet, indel and copy-number mutational signatures to single samples and probabilistically to individual mutations, outperforming four refitting tools on 2,700 synthetic genomes.

A 489,686-cell scRNA-seq atlas of 62 zebrafish stages (3.3–120 hpf) identifies 90 multi-tissue gene programs and markers for pericyte subtypes, intestinal smooth muscle layers, and best4+ enterocytes.

Bounding boxes from a transformer detector let the Segment Anything Model segment cells automatically; one generalist model beats generalist Cellpose and matches inter-annotator agreement across mammalian, yeast and bacterial images.

ENCODE-rE2G, a classifier trained on K562 CRISPR perturbations, maps 13,455,443 enhancer-gene interactions across 352 biosamples, matching or exceeding prior models on CRISPR, eQTL, and GWAS benchmarks.

Autoencoder design plus a roughly 24-hour cell-free screen tested 500 candidates over five rounds, finding 30 functional AMPs; six were broad-spectrum with no resistance emerging in E. coli.

Intronic GBA1 variant rs3115534-G, common in African and African admixed ancestry but rare elsewhere, raises Parkinson's disease risk (OR 1.58) and lowers onset age.

A de novo yeast neochromosome carrying all 275 tRNA genes was built and characterized; it imposes a growth burden, with genome doubling and deletions in haploids possibly reflecting compensation.

Over half of 614 commercial antibodies for 65 human neuroscience proteins failed a knockout-controlled assay, yet most targets had a well-performing antibody; recombinants performed best.

Pairing each residue with its Foldseek 3D-structure token and pre-training an ESM-2-sized model on about 40 million AlphaFold2 structures beat ESM-2 and structure-based baselines on variant-effect and function tasks.

CollecTRI-derived signed regulons, spanning 1,183 transcription factors (TFs) and 45,856 TF-gene links, outperform DoRothEA, ChEA3, RegNetwork and Pathway Commons at identifying perturbed TFs in KnockTF data.

skani estimates average nucleotide identity (ANI) from sparse FracMinHash k-mer chains, staying robust to incomplete, fragmented metagenome-assembled genomes while running over 20 times faster than FastANI.

A Foldseek-based linear-time structural clustering groups 52 million AlphaFold DB representatives into 2.27 million non-singleton clusters; 31% lack known structural or domain annotation.

Theta-burst transcranial ultrasound lowered GABA+ in the posterior cingulate but not dorsal anterior cingulate cortex in 24 adults, while raising resting-state connectivity of both targets.