We evaluated the state of label-free discovery proteomics focusing especially on technological contributions and contributions of naturally occurring differences in protein abundance to the intersample variability in protein abundance estimates in this highly peptide-centric technology. First, the performance of popular quantitative proteomics software, Proteome Discoverer, Scaffold, MaxQuant, and Progenesis QIP, was benchmarked using their default parameters and some modified settings. Beyond this, the intersample variability in protein abundance estimates was decomposed into variability introduced by the entire technology itself and variable protein amounts inherent to individual plants of the Arabidopsis thaliana Col-0 accession. The technical component was considerably higher than the biological intersample variability, suggesting an effect on the degree and validity of reported biological changes in protein abundance. Surprisingly, the biological variability, protein abundance estimates, and protein fold changes were recorded differently by the software used to quantify the proteins, warranting caution in the comparison of discovery proteomics results. As expected, ∼99% of the proteome was invariant in the isogenic plants in the absence of environmental factors; however, few proteins showed substantial quantitative variability. This naturally occurring variation between individual organisms can have an impact on the causality of reported protein fold changes.
Keywords: MaxQuant; Progenesis QIP; biological variability; experimental variability; label-free quantification; natural variability; protein abundance; shotgun proteomics.