HIDDEN: a machine learning method for detection of disease-relevant populations in case-control single-cell transcriptomics data

In case-control single-cell RNA-seq studies, sample-level labels are transferred onto individual cells, labeling all case cells as affected, when in reality only a small fraction of them may actually be perturbed. The standard single-cell analysis approach fails to isolate the affected case cells and their markers when the affected subset is small or the perturbation is mild. To address this, HIDDEN, a computational method, refines case-control labels to accurately reflect the perturbation status of each cell. HIDDEN's superior ability to recover biological signals is demonstrated in simulated datasets and real-world applications to multiple myeloma precursor conditions and a mouse demyelination model. It identifies malignancy in early-stage samples and an endothelial subpopulation involved in blood-brain barrier dysfunction.

Aleksandrina Goeva, Michael-John Dolan, Judy Luu, Eric Garcia, Rebecca Boiarsky, Rajat M. Gupta, Evan Macosko