Head-to-Head CRISPRko CRISPRi Perturb-seq: Differential Expression Analysis results

Results of differential expression analysis from parallel CRISPRko and CRISPRi Perturb-seq screens (Seurat objects available here). Differential expression analysis was performed separately for each sample to inform guide perturbation effects. SCEPTRE low MOI was run in singleton mode on raw sequencing counts to identify genes differentially expressed in cells with each guide relative to cells with intergenic-targeting negative control guides. This method was employed because it uses a permutation-based approach to avoid model misspecification, a relevant concern given the sparsity of single cell transcriptomic data. The n_nonzero_trt_thresh argument of the run_qc method was reduced from its default value of 7 to 0 to enable detection of transcripts with complete loss in perturbed cells. Otherwise, all default pipeline settings were employed. For downstream analysis, the fold change reported from SCEPTRE is incremented prior to manually calculating log10(fold change +0.1) to prevent division by 0 for transcripts absent in perturbed cells.