Human tissues are known to exhibit interindividual variability, but a deeper understanding of the different factors affecting protein expression is necessary to further apply this knowledge. Our goal was to explore the proteomic variability between individuals as well as between healthy and diseased samples, and to test the efficacy of machine learning classifiers. In order to investigate whether disparate proteomics data sets may be combined, we performed a retrospective analysis of proteomics data from 9 different human tissues. These data sets represent several different sample prep methods, mass spectrometry instruments, and tissue health. Using these data, we examined interindividual and intertissue variability in peptide expression, and analyzed the methods required to build accurate tissue classifiers. We also evaluated the limits of tissue classification by downsampling the peptide data to simulate situations where less data is available, such as clinical biopsies, laser capture microdissection or potentially single-cell proteomics. Our findings reveal the strong potential for utilizing proteomics data to build robust tissue classifiers, which has many prospective clinical applications for evaluating the applicability of model clinical systems.
Keywords: bioinformatics; classification; data mining; data reuse; human variability; machine learning.