The front page of AI-driven biology
Thursday, October 15, 2026Thu, Oct 15
Updated 8:01 PM PT
Researchers built REN-former by fine-tuning the Geneformer foundation model on single-cell RNA sequencing from 45 kidney patient samples to predict directional cell shifts from acute kidney injury to chronic kidney disease.
OpenAI has officially backed a package of bipartisan United States congressional legislation aimed at preventing artificial intelligence systems from assisting in the development of biological weapons.
ProtScape combines global protein interaction data, single-cell gene expression, and protein language models to reconstruct context-specific protein interactomes across more than 200 cell types.
LucaCell is a single-cell foundation model that encodes transcriptomic features through pre-trained mRNA nucleotide sequence embeddings rather than fixed gene names, allowing it to translate across species without manual ortholog mapping.
Researchers introduced ICARus, a computational framework that uses protein language model predictions to reduce off-target noise in RNA interference functional screens.
Researchers adapted AgroNT, a transformer-based genomic foundation model, by adding stress-conditioned prompt tokens to pinpoint regulatory regions tied to heat and drought resilience in maize.
Computational biologists introduced DNT, a genomic language model framework that represents paired maternal and paternal chromosome sequences together rather than processing isolated haploid genomes.
Researchers trained a computational pathology foundation model called CRISP on more than 100,000 frozen-tissue sections from 10 medical centers, testing it on 15,000 retrospective slides across 14 tumor types.
Researchers developed ELF, an ensemble learning framework that unifies slide-level representations from five distinct pretrained pathology foundation models.
UniCure links biological and chemical foundation models to forecast drug-induced gene expression changes, aiming to bridge the gap between cancer cell lines and clinical responses.
An independent benchmark of 16 computational models found that methods designed to predict single-cell gene expression directly from standard tissue microscopy images perform poorly across skin disease samples.
Researchers introduced GPN-Star, a genomic foundation model that integrates whole-genome alignments and evolutionary species trees across vertebrate, mammal, and primate lineages.
Researchers combined embeddings from the genomic language model Evo2 with evolutionary ortholog data and protein interaction networks to predict which genes are essential for fungal survival.
InstaNovo-FM is a self-supervised transformer model for bottom-up proteomics trained to interpret peptide mass fragmentation spectra across varying lab instruments and protocols.
AtlasFold is an open-source protein structure prediction model that generates three-dimensional atomic coordinates directly from single protein sequences without requiring evolutionary multiple-sequence alignments.