The label‐free protein quantifications (precursor areas) for the colorectal tumours (Zhang et al, 2014 (link)) and the tandem mass tag (TMT) protein intensities for the breast and colorectal cancer cell lines (Lawrence et al, 2015 (link); Lapek et al, 2017 (link); Roumeliotis et al, 2017 (link)) were pre‐processed and transformed to log2 fold‐changes as described previously (Sousa et al, 2019 (link)). For the brain (Petralia et al, 2020 (link)), lung (Gillette et al, 2020 (link)) and stomach (Mun et al, 2019 (link)) cancers, the sample replicates were combined by averaging the log2 fold‐change values of each protein. After that, we removed six outlier samples from colorectal cancer with an absolute median log2 fold‐change distribution higher than 1 (2‐fold). Altogether, we assembled a matrix with 14,742 proteins and 1,266 samples (1,170 cancer samples and 96 cell lines) belonging to nine different tissues. This matrix contained 9,941,918 protein measures (8,721,454 missing values) and 5,052 proteins quantified in at least 80% of the samples.
Free full text: Click here
