Fig. 1: In vitro RAS signature derivation.
From: RAS oncogenic activity predicts response to chemotherapy and outcome in lung adenocarcinoma
a Contingency tables showing the number of RAS pathway wild-type and KRAS mutant cell lines per RAS-high and RAS-low groups for each signature. RAS pathway wild-type cell lines are those with no oncogenic mutation in any RAS pathway member. RAS-high cell counts are shown in red, RAS-low in blue. The boxplots show the RI distributions for the RAS-high and low groups where the box shows the median and IQR, the whiskers indicate ± 1.5 x IQR, outliers lie outside this range. RAS addiction is presented here as a control signature (RAS pathway n = 115, KRASG13D134 n = 137, HRAS n = 106, MSigDB n = 124, KrasLA n = 98, RAS addiction n = 124; Chisq test ****P ≤ 0.0001, ***P ≤ 0.001, **P ≤ 0.01, n.s.= P > 0.05). b Heatmap showing our RAS84 meta-signature genes mapped to filtered (see “Method”) CCLE lung cell line data. Cell lines are shown as rows, genes as columns. Groupings of high, medium and low RAS activity are shown as separate clusters, KRAS mutational status is indicated in dark red on the right and parent signature gene membership is indicated in grey at the bottom of the map. c Contingency tables showing the number of RAS pathway wild-type and KRAS mutant cell lines per RAS-high and RAS-low groups for RAS84. RAS pathway wild-type cell lines are those with no oncogenic mutation in any RAS pathway member. RAS-high cell counts are shown in red, RAS-low in blue. The boxplots show the RI distributions for the RAS-high and low groups where the box shows the median and IQR, the whiskers indicate ± 1.5 x IQR, outliers lie outside this range. N = 120, Chisq test ****P ≤ 0.0001, ***P ≤ 0.001, **P ≤ 0.01, n.s. = P > 0.05. d Log-likelihood values from a GLM fit (family = binomial) of KRAS mutation status across the three RAS activity groups for each of the signatures. RAS84 is the best-performing signature at segregating KRAS mutants across the RAS activity groups.
