Computational identification of clonal cells in single-cell CRISPR screens

Single-cell CRISPR screens are powerful tools to understand genome function by linking genetic perturbations to transcriptome-wide phenotypes. However, since few cells can be affordably sequenced in these screens, biased sampling of cells could affect data interpretation. One potential source of biased sampling is clonal cell expansion. Here, we present a computational pipeline for clonal cell identification in single cell screens using multiplexed sgRNA barcodes. We find that the cells in each clone share transcriptional similarities and we infer the segmental copy number changes in clonal cells. These analyses suggest that clones are genetically distinct. Finally, we show that the transcriptional similarities of clonally expanded cells contribute to false positives in single-cell CRISPR screens. As a result, experimental conditions that reduce clonal expansion or computational filtering of clonal cells will improve the reliability of single-cell CRISPR screens. We performed single-cell CRISPR screen on 2045 enhancers in MDA-MB-231 cells. We used 10X genomics 3’ V3 kit and Cell Hashing staining to prepare the single-cell transcriptome libraries. The sgRNA is expressed by using the CROP-seq design, and a sgRNA enrichment library was also prepared for each single-cell library. In total, we performed six 10X runs, with a total number of 50000 singlets. For the processed data, we uploaded the gene expression matrix (HDF5 format) from 10X cellranger pipeline in together with the sgRNA information and Cell Hashing information of each cell.