Evaluating Mutation Effects on Kinase and TF Activities

The effects of mutations on the kinase and TF activities were assessed by associating the activity of a given protein with the mutational status of the same protein or other proteins it might interact throughout the cellular regulatory networks. First, we built a binary mutation matrix M where the index M_ij corresponds to 1 if the sample i has a mutation in gene j and 0 otherwise. To do that, we selected the mutations classified as frameshift and in‐frame Indels, missense, nonsense and stop codon loss. Given the proteins X and Y, the association between the activity of Y (Y_act) and the mutational status of X (X_mut) was assessed across samples by fitting a linear model that took into account possible confounding effects: $${\mathrm{Y}}_{\mathrm{act}}={\mathbf{\beta }}_0+{\mathbf{\beta }}_1\text{Study}+{\mathbf{\beta }}_2{\mathrm{X}}_{\mathrm{mut}}+\mathrm{\varepsilon }$$ where Y_act represents the activity of protein Y, β₀ the intercept, β₁ the regression coefficient for the covariate experimental study, β₂ the regression coefficient for the mutational status of X and ɛ the noise term. This model was applied to assess the effect of X_mut on the activity of the same protein (X_act ∼ X_mut) and on the activity of other proteins (Y_act ∼ X_mut). The P‐values from the coefficients of X_mut (β₂) were calculated using the t‐statistic over a Student's t‐distribution and adjusted for false discovery rate (FDR) using the Benjamini–Hochberg method. The linear models and respective statistics were calculated using the lm and p.adjust R functions.
The associations were performed with the genes mutated in more than 20 samples and with the protein activities estimated in at least 10 samples. An association between a pair Y_act ∼ X_mut or X_act ∼ X_mut was performed if X_mut was mutated in at least five of all the samples in the pair. Regarding the Y_act ∼ X_mut associations, we tested 520,938 pairs between 208 kinases and 3,590 genes and 1,048,216 pairs between 292 TFs and 3,590 genes. In relation to the X_act ∼ X_mut associations, we tested 40 pairs and 64 pairs with the kinases and TFs, respectively.