Generalized inference of noise-corrected state variability in single cells

stochprofIC infers noise-corrected single-cell state variability from mini-bulk data, enabling deconvolution of heterogeneous nucleocytoplasmic trafficking states in mouse brains and human breast cancers.

• Why it matters: Single-cell omics are limited by detection noise and weaker technical controls than mini-bulk, making it hard to integrate split-pool controls and reliably infer underlying cell-state variability.
• What they did: The authors developed stochprofIC, a distributional-regression algorithm that models data, noise, and covariates, and validated it on synthetic datasets and by comparing against an adjusted-variance alternative using mini-bulk mouse brain and human breast cancer data.
• The result: stochprofIC’s noise-batch corrections support inference of the single-cell populations composing mini-bulk, repeatedly validated by RNA FISH, and its outputs distinguish trafficking states; MAP3K11-high state inhibition reduces apoptosis in primary breast-cancer organoids.

AI summary of the abstract · may contain errors · not medical advice · read the paper.