Top 40
View all ›BACKFILLED Code and industry signals are partial for backfilled days.
ML / AI methods
Biological priors constrain deep models of spatial gene expression and histology
Spatial omics and single-cell models increasingly encode ontology, marker, and pathway knowledge as inductive structure rather than new architectures.
Single-cell
New tools for cell type annotation across single-cell and spatial data
Annotation consistency is being addressed through atlas construction, summary-statistic label transfer, and LLM-based ontology harmonization.
ML / AI methods
Pretrained biomolecular language models: what their frozen representations encode
Recent preprints and publications probe what protein and genomic language model embeddings encode, beyond predictive accuracy alone.
Structural bio & protein design
Deep learning models predict protein stability and binding effects of mutations
Preprints and publications now pair calibrated protein language models, physics hybrids, and inverse-folding models with variant stability, binding, and…
Systems biology
Single-cell and network models map non-coding regulatory variants to gene activity
Context-aware regulatory networks and single-cell enhancer-gene models now link non-coding variants to disease and chromatin state.