Top 40
View all ›BACKFILLED Code and industry signals are partial for backfilled days.
ML / AI methods
Histology-to-spatial-transcriptomics prediction moves to diffusion over gene programs
Histology-conditioned models are moving beyond per-gene regression toward program-level targets, cross-modal alignment, and multi-scale biological priors.
Transcriptomics
Learned mappings infer unmeasured transcriptomic states from partial observations
Recent frameworks address missing modalities and context-dependent expression in spatial and live-cell data using learned mappings and explicit uncertainty…
ML / AI methods
Perturbation data now trains LLM reasoning and learned predictors of cellular response
Perturbation atlases now serve as RL environments for LLM biological reasoning, alongside learned models of response and sequence design.
Systems biology
Synaptic delays, pulsed control, and STN spike models shape oscillatory synchrony
Model dependence in E-I and basal ganglia networks now bears directly on beta and synchrony mechanisms relevant to Parkinson's disease.
Single-cell
Model-based methods reshape batch correction, gene selection and cell-type deconvolution
Recent preprints and publications benchmark model-based tools for transcriptome extrapolation, gene selection, batch correction and deconvolution.