Top 40
View all ›BACKFILLED Code and industry signals are partial for backfilled days.
Single-cell
Generative and graph models predict single-cell perturbation and regulatory states
Recent preprints pair flow-matching, Schrödinger bridge, and graph-learning models with Perturb-seq and multiome data to predict perturbation effects.
ML / AI methods
Deep learning bioactivity models are scrutinized under leakage control and uncertainty
Rigorous benchmarking of leakage, calibration, and generalization now accompanies new models for affinity, permeability, and ADME prediction.
Structural bio & protein design
Ensemble sampling and structural priors sharpen cryptic pocket and mutation prediction
Kryptix-1 raises cryptic pocket success to 80% versus 20% static, while mutation and fitness models extend structural context to affinity and stability.
Single-cell
Single-nucleus multiomics resolve cell-type gene regulation across tissues and species
Donor-scale and cross-species single-nucleus datasets now link cell-type regulation to brain traits, disease variation, and conservation.
Structural bio & protein design
Integrative and simulation methods map RNA and protein conformational landscapes
Combined smFRET, MD, Markov models and elastic-network sampling now resolve RNA and protein conformational ensembles at scale.