Fig. 4
From: Genotype-free demultiplexing of pooled single-cell RNA-seq
The overall pipeline of scSplit tool. a SNV identified based on reads from all cells which have similar or different genotypes. b Alternative and reference allele count matrices built from each read in the pooled-sequenced BAM at the identified informative SNVs. c Initial allele fraction model constructed from the initial cell seeds and their allele counts. d Expectation-maximization process to find the most optimized allele fraction model, based on which the cells are assigned to clusters. e Presence/Absence matrix of alternative alleles generated from the cell assignments. f Minimum set of distinguishing variants found to be used to map clusters with samples