Mass Spectrometry-Based Proteomics of Minor Species in the Bulk: Questions to Raise with Respect to the Untargeted Analysis of Viral Proteins in Human Tissue

SIMPLE SUMMARY: Proteomics is a booming field in life sciences and, increasingly, is not only based on mass spectrometry (MS), but also on multiplex bead or aptamer assays and proximity extension assays. The use of untargeted MS generates big data sets in little time, but it also has limitations. Here, we discuss the plausibility of screening proteomic shotgun MS raw data for viral proteins in human gastric biopsies. Though this is technically possible and, thus, appealing to researchers, low-abundant proteins of guest species are barely present at suitable concentrations for measurement. Still, the processing algorithms will return hits due to chance assignments of low-quality spectra, but these should be red-flagged. A sanity check needs to establish whether or not certain proteins can be available at a sufficient concentration for measurement in the sample at all. Not every possible analysis is, thus, sensible. Even though both instrumentation and bioinformatic processing are continuously improving, a quality control of the data output will always be advisable. This paper uses practical examples to explain difficulties in spectral assignment leading to false-positive protein matches and is, thus, a tutorial for laymen and novices in MS-based proteomics. ABSTRACT: (1) Background: Untargeted mass spectrometry (MS)-based proteomic analysis is highly amenable to automation. Software algorithms translate raw spectral data into protein information obtained by a comparison to sequence databases. However, the technology has limitations, especially for analytes measured at the limit of detection. In a protein expression study of human gastric biopsies, the question arose whether or not it is possible, as well as sensible, to search for viral proteins in addition to those from the human host. (2) Methods: Experimental data-independent MS data were analyzed using protein sequences for oncoviruses, and BLAST analyses were performed to elucidate the level of sequence homology to host proteins. (3) Results: About one hundred viral proteins were assigned, but there was also up to 43% sequence homology to human proteins. (4) Conclusions: There are at least two reasons why the matches to viral proteins should be used with care. First, it is not plausible that large amounts of viral proteins should be present in human gastric biopsies, so the spectral quality of the peptides derived from viral proteins is likely low. As a consequence, the number of false assignments is high. Second, homologous peptides found both in human and virus proteomes contribute to matching errors. Thus, though shotgun proteomics raw data can technically be analyzed using any database, meaningful results cannot be always expected and a sanity check must be performed. Both instrumentation and bioinformatic processing in MS-based proteomics are continuously improving at lowering the limit of detection even further. Nevertheless, data output should always be controlled in order to avoid the over-interpretation of results.