Fig. 1. Combinatorial CRISPR screen for cancer targets.

A Schematic of the genes and gene pairs included in the paired gRNA library. The number of gene pairs for which the minimum guide number could be identified with our gRNA selection criteria is shown. The number in brackets is the total number of pairs in each category. B Design of the multiplex gRNA construct. To assess single-guide activity, a non-targeting gRNA (Fluc) was placed under the hU6 promoter and the gRNA against the candidate gene was placed under the sU6 promoter. To assess paired gene activity, all five gRNAs for one gene were under the hU6 promoter, whereas the guides targeting the other gene were under the sU6 promoter. C Derivation of residual values. The predicted log₂-fold change (FC) of the paired construct was calculated from the sum of the growth phenotypes of individual guides and the observed (experimentally determined) log₂FC was the actual behaviour of the pair in the screen. The behaviour of the population was modelled using Loess smoothing (blue line). The residual was calculated as the vertical difference of any given point to the blue line. The more negative the residual, the more lethal the pair in the screen. The behaviour of the gRNA pairs for the nine synthetic lethal interactions common to all cell lines is shown in the A375 cell line. D Overlap between the synthetic lethal pairs identified in the screen in each of the three cell lines at Day 28. Significant gene pairs were defined as having a significantly more lethal pairwise effect on cellular fitness than expected from the effect of each individual gene (see Methods). Pairs where one of the genes was found to be lethal in isolation at Day 14 were excluded. E Percentage DNA sequence similarity between paralogues in our screen, identified as synthetic lethal in ≥1 cell line (Ever SL) vs zero cell lines (Never SL). Error bars show mean and standard deviation (SD) with each data point representing the average sequence similarity between paralogue pairs (see Source Data). The p value was determined using the two-tailed Mann–Whitney test.