Fig. 6: Fitting regression models to represent locus efficiency.
a A comparison of two loci within lung squamous cell tumors (N = 126): sum of locus RNA expression within these samples (blue bar) scaled to the fitted efficiency value (orange bar), sum of locus TRT counts within these samples (green bar) (see “Methods” for calculation of efficiency). b Linear regression of coefficients assigned to individual loci (y-axis) vs. in vitro activity as measured by Brouha et al.24 (x-axis), reported as a relative percentage of measured L1RP activity. N = 76 loci (subset of 156 that were measured in vitro24), R = 0.63, p = 9.3 × 10−10, Pearson correlation (using the exact distribution, as calculated by the scipy.stats.pearsonr function in python). Error bars represent 95% confidence intervals around assigned coefficient. Red box indicates region shown in (c). c Linear regression, removing 4 high-activity loci (“hot” L1s from Brouha et al.24). N = 72 loci, R = 0.42, p = 2 × 10−4, Pearson correlation (exact distribution). Similarly, error bars represent 95% confidence intervals around assigned coefficient.
