Top 40
View all ›BACKFILLED Code and industry signals are partial for backfilled days.
ML / AI methods
Benchmarks reveal limits of deep learning for protein and ligand prediction
Protein language models and structure-aware networks are now evaluated under leakage control, uncertainty calibration, and closed-loop design.
Structural bio & protein design
Conformational ensembles and learned priors sharpen pocket and interface prediction
Learned priors and conformational sampling now target cryptic pockets, interface energetics, and ligand-induced state changes.
Single-cell
Benchmarks and single-cell atlases test expression and regulatory inference methods
Single-cell foundation models and in silico perturbation tools are now being assessed against aging, cardiac, and multi-tissue reference data.
Genomics
Chromosome-scale assemblies reveal repeat expansion and host-integrated bacterial DNA
Long-read and near-T2T assemblies now resolve host-integrated bacterial chromosomes, repeat-driven size variation, and graph-based structural variation.
Single-cell
Structured models infer single-cell dynamics and metabolic states from transcriptomic data
Structured latent models now embed growth laws, metabolic topology, RNA velocity geometry and lineage trees to make single-cell inference interpretable.