Fig. 1
Baseline reproducibility between the Project Achilles and COLT-Cancer genome-wide shRNA screens. a Overlap in shRNAs, target genes, and cell lines screened in the Achilles and COLT-Cancer projects. Based on sequence identity, we found 46,474 shRNAs were commonly profiled in Achilles 2.4 and COLT-Cancer (top Venn diagram); based on The RNAi Consortium clone identifier, 50,966 shRNAs were commonly profiled in Achilles 2.0 and COLT-Cancer (bottom Venn diagram). b An example scatterplot of shRNA essentiality scores (shES) in Achilles 2.4 and COLT-Cancer studies across overlapping shRNAs profiled in the CAL51 cell line. The between-study consistency was assessed using Spearman rank correlation (ρ). The red and blue dots highlight those shRNAs having strong and weak seed pairing stability (SPS), respectively (see “Methods” for detailed description). c Inter-study correlation (ρ) for shES across matched cell lines between Achilles 2.4, Achilles 2.0, and COLT-Cancer studies. The grey dashed line indicates average correlation (ρ = 0.38) over the 13 cell lines between Achilles 2.0 and COLT-Cancer; the black dashed line average correlation (ρ = 0.57) over the 23 cell lines between Achilles 2.4 and COLT-Cancer; and the red dashed line average correlation (ρ = 0.61) over the 17 high data quality cell lines between Achilles 2.4 and COLT-Cancer (asterisks indicate cell lines with low replicate correlation ρrep < 0.5). d Intra-study correlation (ρ) for shES between Achilles 2.0 and 2.4. The black dashed line indicates average correlation over the 12 matching cell lines (ρ = 0.70). The baseline consistency between the two screens was moderate based on the shES provided in the two studies; the Achilles study scores the shRNA essentiality using normalized fold changes between initial and final time points, averaged over the replicates, whereas the COLT-cancer study uses the so-called shARP score, which is estimated as the ratio of change in expression intensity of the shRNAs over population doublings