Top 40
View all ›BACKFILLED Code and industry signals are partial for backfilled days.
ML / AI methods
Biological foundation models are benchmarked for efficiency, fidelity and transfer
Recent preprints and papers scrutinize scaling, multimodal pretraining, and quantization limits in biological sequence and single-cell models.
ML / AI methods
Benchmarks show in silico perturbation and graph models depend on method and cell context
Preprints now benchmark in silico perturbation methods and extend graph and attention models to cell-line, tissue and clinical biomarker tasks.
Transcriptomics
Statistical methods for separating biological and technical variation in omics data
Recent work introduces methods that separate known covariates, technical artifacts, and rhythmicity from biological variation in omics data.
Structural bio & protein design
Structure prediction models are benchmarked for antibody and peptide binder design
Blinded prospective and co-folding benchmarks now test AI-designed binders against experimental affinity and pose accuracy.
Genomics
Benchmarking refines variant detection across artifacts, pangenomes, and ancient DNA
Recent work benchmarks variant detection and correction against gold standards across FFPE artifacts, pangenome SVs, and mtDNA heteroplasmy.