DevKidCC allows for robust classification and direct comparisons of kidney organoid datasets

Abstract

Kidney organoids provide a valuable resource to understand kidney development and disease. Clustering algorithms and marker genes fail to accurately and robustly classify cellular identity between human pluripotent stem cell (hPSC)-derived organoid datasets. Here we present a new method able to accurately classify kidney cell subtypes, a hierarchical machine learning model trained using comprehensive reference data from single cell RNA-sequencing of human fetal kidney (HFK). We demonstrate the tools (DevKidCC) performance by application to all published kidney organoid datasets and a novel dataset. DevKidCC is available on Github and can be used on any kidney single cell RNA-sequence data.