Top 40
View all ›BACKFILLED Code and industry signals are partial for backfilled days.
Single-cell
Learned models impute, translate and infer dynamics in single-cell and spatial data
Foundation models, generative flow matching and optimal transport now reach imputation, counterfactuals and trajectory inference across modalities.
ML / AI methods
Foundation models unify histology and spatial transcriptomics for tissue analysis
Benchmarks now show frozen genomic and single-cell foundation model rankings are unstable and that reported gains depend on the evaluation protocol.
Epigenomics
Computational models link chromatin looping, methylation and CTCF to regulatory output
Integrative models now resolve loop strength, Tn5 bias, and enhancer identity from multi-omic chromatin data across cell types and individuals.
ML / AI methods
Benchmarked ML models predict protein properties, with stated generalization limits
New preprints pair predictive gains for enzymes, protein stability and interactions with explicit benchmark-based limits.
Transcriptomics
Computational methods extend spatial transcriptomics to niches and chromatin
Recent work pairs graph, variational, and Bayesian learning with spatial and single-cell transcriptomic data for niche, regulator, and panel inference.