Skip to main content
Fig. 3 | Genome Medicine

Fig. 3

From: Seed-effect modeling improves the consistency of genome-wide loss-of-function screens and identifies synthetic lethal vulnerabilities in cancer cells

Fig. 3

Reproducibility of the seed essentiality scores with increasing shRNA family size of seed sequences. a Average rank correlation (ρ), with standard error of mean over the 17 high data quality cell lines (error bars), calculated based on seed essentiality score (SeedES) as a function of shRNA family size (x-axis). shRNAs sharing the same seed sequence belong to the same shRNA family. The red trace indicates the observed correlation based on seed region. The blue trace indicates the correlation based on heptamer12–18ES for positions 12–18. The black trace indicates correlations based on 1000 permutations over the seed—shRNA mapping (see “Methods” for details). The gray dotted line indicates the intra-study correlation for shES between Achilles 2.0 and 2.4 (ρ = 0.70). SeedES-based inter-study correlation reached its maximum at family size of 14 (ρ = 0.77), suggesting that the consistency between the studies increases when off-target effects are more accurately estimated using larger family size. Asterisks indicate statistically significant differences in correlations (p < 0.05, paired t-test), and their colors indicate the distribution against which the comparison was done. b Density distribution of shRNA family size of overlapping shRNAs profiled in the two shRNA screens. Family sizes with greater than 50 unique seeds were considered in the analysis. The gray dotted line indicates the shRNA family of size 5

Back to article page