Fig. 4

Heat map of Spearman rank correlation coefficients between efficiency scores and datasets. For each dataset, the experimental system is indicated by a species icon or cell type. Number of guides tested are shown in parentheses. Scores are shown along the horizontal axis, datasets on the vertical. Correlations of an algorithm against its own training dataset are shown in grey as they are likely to be overestimated due to overfitting. The datasets Wang/Xu HL60 and Koike-Yusa/Xu on mouse embryonic stem cells are originally from Wang et al. [38] and Koike-Yusa et al. [54] but were used as processed by Xu et al. [23]. From the dataset by Hart et al. [41], only the cell line Hct116/repeat2 was used, as it gave the highest correlation value; for this study efficiency was averaged over all time points. Data on human cell lines for the two datasets by Doench et al. [8, 34] are not shown here but gave an almost identical correlation profile. All data points are shown as scatter plots in Additional file 12; for assay background information on the datasets see Additional file 8: Table S4