Fig. 4.

Co-dependencies in RNAi, CRISPR, and integrated datasets. a Co-dependencies of each query gene (high-confidence dependencies with at least 3 dependent cell lines in either CRISPR or RNAi dataset) were tested for enrichment of previously established related genes (CORUM, PPI, KEGG, paralogs). Query genes that were classified as CRISPR pan-dependencies are separated from the others. b Number of query genes for which one of the gene’s related priors is among the query gene’s top 1, 5, 10, or 100 ranked co-dependencies. Query genes are separated by CRISPR pan-dependency status as in part b. c Examples of local co-dependency networks for query genes, RPS21 (pan-dependency) and MED16 (selective dependency), constructed from CRISPR, RNAi, and integrated CRISPR-RNAi gene similarity networks. Only the top 10 co-dependencies of the query gene are included as vertices, but a directional link is shown between any two vertices if the target is a top 10 co-dependency of the source. Size of a vertex corresponds to the number of connections between the vertex and other gene vertices in the network. Edge weights are scaled respective to the full similarity matrix for each dataset