Here is a method description from from my previous experience (see this notebook):
To assess pathway-level enrichment of gene phenotypes in the CRISPRi screen, we used blitzGSEA28, a Python package for the computation of Gene Set Enrichment Analysis (GSEA) (https://github.com/MaayanLab/blitzgsea). We obtained gene ontology (GO)30 gene sets from MSigDB30,69,70 (version 7.4.) and then conducted two separate analyses: (1) To identify smaller29, focused pathways associated with drug sensitivity or resistance, we performed GSEA analysis on genes ranked by ρ phenotype and defined minimum and maximum thresholds for gene set size when running the ‘gsea‘ function (‘min_size=15‘ and ‘max_size=150‘). Thus, positive normalized enrichment scores (NES) corresponded to gene sets enriched among positive ρ phenotypes (i.e., resistance phenotypes) and negative NES corresponded to gene sets enriched among negative ρ phenotypes (i.e., sensitivity phenotypes). (2) To identify broader pathways associated with drug response irrespective of ρ phenotype direction, we performed GSEA analysis on genes ranked by 1 – Mann-Whitney p-value (calculated for each ρ phenotype as above) and set a minimum threshold for gene set size (i.e., ‘min_size=200‘). Then we only report positive normalized enrichment scores (NES) that represent gene sets enriched by more significant Mann-Whitney p-value for the gene set’s ρ phenotype.
blitzGSEA
Here is a method description from from my previous experience (see this notebook):