Fig. 3.

Comparisons of counts and singscores within samples from NanoString assay. A Linear regression between 12 NanoString repeats from 5 samples. The repeats were derived from samples from the same patients (Pt1-5). Linear regression of raw count data (top lane); normalised count data (middle lane); singscores using the “No stable gene” method (bottom lane). B Boxplot of Spearman correlation (r) and Linear regression coefficients, including the r², slope and intercept, between singscores of 81 signatures derived from raw and normalised NanoString counts using “No stable gene” method of 165 samples