Ambient RNA analysis reveals misinterpreted and masked cell types in brain single-nuclei datasets

Abstract

Ambient RNA contamination in single-cell RNA sequencing (RNA-seq) is a significant problem, but its consequences are poorly understood. Here, we show that ambient RNAs in brain single-nuclei RNA-seq can have a nuclear or extra-nuclear origin with distinct gene set signatures. Both ambient RNA signatures are predominantly neuronal and we find that some previously annotated neuronal cell types are distinguished by ambient RNA. Strikingly, we also detect pervasive neuronal ambient RNA contamination in all glial cell types unless glia and neurons are physically separated prior to sequencing. We demonstrate that this contamination can be removed in silico. We also show that previous annotations of immature oligodendrocytes are likely glial cells contaminated with ambient RNAs. After ambient RNA removal, we detect extremely rare, committed oligodendrocyte progenitor cells, which were infrequently annotated in previous adult human brain datasets. Together, these results provide an in-depth analysis of ambient RNA contamination in brain single-cell datasets.