Top 40
View all ›BACKFILLED Code and industry signals are partial for backfilled days.
ML / AI methods
Evaluation design, not model choice, shapes reported agreement in biomedical ML
Imputation, harmonization, and label-set choices now change reported performance across ICU, transcriptomic, multi-omics, and pathology benchmarks.
ML / AI methods
Low-variance rotational modes and ground-truth benchmarks for modeling neural dynamics
Recent preprints and a published toolkit push neural dynamics models toward interpretable, ground-truth-validated analysis of brain-wide and behavioral data.
Genomics
Chromosome-scale genome assemblies enable locus-level repeat and expression analysis
Chromosome-level and haplotype-resolved assemblies across diverse taxa now serve as references for repeat, transcript, and pangenome analysis.
Genomics
Mutation signatures and benchmarks resolve cancer genome history and structure
Standardized somatic variant benchmarks and amplicon-resolved workflows now let cancer genome mutation histories be inferred with benchmarked accuracy.
Tools & infrastructure
Benchmarks reveal tool-dependent accuracy in nanopore and metagenomic profiling
Benchmarks across read filtering, viral detection, binning, 16S classification and strain profiling show tool choice governs accuracy in microbial genomics.