Top 40
View all ›BACKFILLED Code and industry signals are partial for backfilled days.
ML / AI methods
AI molecular design is increasingly judged by validation, dataset quality and ADMET
Generative and property-prediction models are now being stress-tested with realistic splits, dynamics-aware scoring and dataset-quality metrics.
Structural bio & protein design
Binder design shifts from generation toward computational evaluation and filtering
Prospective blinded challenges and benchmarks now test whether AI-designed binders hold up against experimental affinity and developability.
ML / AI methods
Pretraining and long-context designs reshape biological foundation model evaluation
Benchmarks of biological foundation models now test data modality, context length, and quantization tradeoffs rather than parameter scale alone.
Tools & infrastructure
LLMs and explicit ontologies convert genomic records into queryable knowledge
Recent LLM agent workflows now annotate GEO and PubMed text at scale, but validation against external resources remains uneven.
Structural bio & protein design
Geometric and sequence models predict protein binding, docking and interface quality
Recent preprints and papers apply sheaf, geometric, state-space and graph learning to protein-ligand and protein-RNA binding tasks.