PT  - JOURNAL ARTICLE
AU  - Erica AK DePasquale
AU  - Kyle Ferchen
AU  - Stuart Hay
AU  - H. Leighton Grimes
AU  - Nathan Salomonis
TI  - cellHarmony: Cell-level matching and comparison of single-cell transcriptomes
AID  - 10.1101/412080
DP  - 2018 Jan 01
TA  - bioRxiv
PG  - 412080
4099  - http://biorxiv.org/content/early/2018/12/17/412080.short
4100  - http://biorxiv.org/content/early/2018/12/17/412080.full
AB  - To understand the molecular etiology of human disease, precision analyses of individual cell populations and their molecular alternations are desperately needed. Single-cell genomics represents an ideal platform to enable to the quantification of specific cell types, the discovery of transcriptional cell states and underlying molecular differences that can be compared across specimens. We present a new computational approach called cellHarmony, to consistently classify individual cells from a query (i.e., mutant) against a reference (i.e., wild-type) dataset to discover crucial differences in discrete or transitional cell-populations. CellHarmony performs a supervised classification of new scRNA-Seq data against a priori delineated cell populations and associated genes to visualize the combined datasets and derive consistent annotations in a platform-independent manner. Such analyses enable the comparison of results from distinct single-cell platforms against well-curated references or against orthogonal profiles from a related experiment. In addition, cellHarmony produces differential expression results from non-confounded aligned cell populations to explore the impact of chemical, genetic, environmental and temporal perturbations. This approach works seamlessly with the unsupervised classification and annotation of cell-states using the software ICGS in AltAnalyze. Using cellHarmony, we demonstrate novel molecular and population insights in scRNA-Seq data from models of Acute Myeloid Leukemia, across technological platforms and using references derived from the Human Cell Atlas project.