Fig. 2

Transcriptomic changes of genotyped cells accurately classify missense mutations into three functional categories. a Correlations of differential gene expression of each barcode to cells with WT genotypes, including non-targeting gRNAs. For each barcode, the consequence and predicted consequence of homozygous mutations are shown. The barcodes fall into two groups: one consisting mainly of stop codons, splice variants, and missense mutations and a second one containing many WT barcodes. b Diffusion map showing a low-dimensional representation to identify the main directions of variation. c The first diffusion component (diffusion score) acts as a measure of loss of function with a high diffusion score for homozygous stop codons and a low diffusion score for WT and homozygous synonymous mutations. d Demonstration of low false-negative and false-positive rates for calling edits in barcode groups. Diffusion score for barcodes called homozygous stop codons and WT. e Possible phenotypic consequences of small differences in editing. Barcodes with the same gRNA but different edits (heterozygous versus homozygous, one edit versus two consecutive edits). The position is the editing position on chr1. Note that editing may be different for the three alleles of the same cells. f Transcriptomic heterogeneity of homozygous missense variants. Density plots for the diffusion scores of all barcodes with homozygous missense variants, including variants with low impact (low diffusion score indicating no LoF-benign), intermediate diffusion scores (indicating separation of function (SoF)), and high impact (high-score missense) mutations. Boxed barcodes highlight variants with intermediate diffusion scores, characterized by lower FACS scores and higher proliferation scores (SoF)