Top 40
View all ›BACKFILLED Code and industry signals are partial for backfilled days.
ML / AI methods
Learned models predict perturbation responses and design molecules and binders
Recent preprints extend Schrödinger bridges, sparse autoencoders, reinforcement learning, and knowledge graphs to perturbation and molecular design tasks.
Single-cell
Injecting LLM gene priors and regulatory scaffolds into single-cell and spatial models
Large single-cell pretraining corpora and multimodal assays now make built-in biological priors a practical route to interpretable predictions.
Structural bio & protein design
Learned models address protein dynamics, property prediction and engineering benchmarks
Recent preprints and publications pair learned potentials and trajectory models with explicit evaluation on conformational and engineering tasks.
Structural bio & protein design
Deep models couple structure-level objectives with antibody, peptide, and ligand design
June 2026 preprints pair predicted-structure supervision with generative and ranking models for binder design, screening, and kinetic prediction.
Epigenomics
Machine-learned models refine chromatin state and regulatory element annotation
Large reference epigenome collections such as IHEC now support continuous, cross-cell-type chromatin state annotation alongside sequence-based models.