Top 40
View all ›BACKFILLED Code and industry signals are partial for backfilled days.
Transcriptomics
Histology-conditioned models infer spatial gene expression and tissue niches
Pretraining at scale links H&E morphology to spatial gene expression, with TissueFormer on 17M pairs and TERRA on 112M cells.
Single-cell
Computational methods refine single-cell annotation, copy number and regulatory inference
Recent preprints and papers add reference-free, summary-level and benchmarked methods for single-cell annotation, copy number and regulation.
ML / AI methods
Learned models predict unseen perturbation responses across transcriptomic settings
Held-out evaluation of foundation models and compact linear predictors now tests whether learned transcriptomic representations generalize beyond training…
ML / AI methods
Learned models predict protein stability, peptide-MHC binding, and antisense activity
Benchmarks and calibrated predictors for stability, binding, and LLM epitope reasoning now emphasize leakage-controlled, held-out evaluation.
Transcriptomics
Statistical frameworks for spatial, single-cell and chromatin genomics data
Spatial transcriptomics tools now model cell-cell communication, neighborhood structure and spatial colocalization with explicit uncertainty handling.