Top 40
View all ›BACKFILLED Code and industry signals are partial for backfilled days.
ML / AI methods
Validated omics methods: ground-truth benchmarks, simulators and uncertainty estimates
Ground-truth simulators and benchmarks now let omics tools be judged against defined references rather than assumed performance.
Single-cell
Predicting spatial expression from H&E morphology via generative and contrastive models
Recent preprints test whether H&E morphology can recover spatial transcriptomes, with molecular supervision from paired data or single-cell references.
ML / AI methods
ML models decode binding and sequence-function rules in peptides, TCRs, and proteins
Recent benchmarks and models test whether learned sequence-to-function maps hold for unseen peptides, modifications, and designed sequences.
Single-cell
Linking dark RNA isoforms and splicing variants to proteomic and genomic evidence
Long-read RNA-seq, mass spectrometry, and structure prediction are now coupled to resolve non-canonical isoforms and benchmark fusion detection.
Epigenomics
Computational models dissect TAD boundary heterogeneity and pangenome variation
Explicit graph and statistical models are now applied to human pangenome variant catalogs and TAD boundary classes at genome scale.