The global experimental design is described in Figure S9 in Additional file 1. In the first stage of the study, we used the four CAL samples to estimate the performance of the assay. The performance was evaluated by successively using one CAL sample to calibrate the algorithm and a different one as a test sample, performing all possible permutations between the four CAL samples as calibration and test samples (Table S5 in Additional file 2). The second stage of the study applied the algorithm to cancer samples (primary and xenografts) using the CAL samples sequenced on the same run for calibration.
