Comparison of Tools for the Intensity-Based Quantification of Proteomic Data

Zack Heins (Mentor: Dr. Renee Olano, Bioinformatics Specialist, National Institute of Allergy and Infectious Diseases, NIH)

September 1st, 2015, 2:40-3:00pm, Room 1300, Harris Building

Mass Spectrometry-based proteomics has two main challenges – the first involving the process of turning data into confident peptide identifications and protein assignments, and the second being the determination of relative abundance as a form of quantification. One of the major approaches to the latter problem, in both label-free and labeled quantification, is to use chromatographic area-under-the-curve calculations. Unfortunately, in practice the latter is usually an add-on to a program that does the former, and there is little information provided on the accuracy of the algorithms incorporated into the tools. Preliminary analysis of a range of available quantification options, as implemented in a variety of proteomic analysis programs, indicates that many suffer from an overestimation of differences in peptide and protein levels.

In this analysis, three types of samples – two single protein samples with multiple acquisitions at a range of concentrations, and a dynamic range proteomic standard – are processed using XCMS, MZmine, Skyline, Proteome Discoverer, and ProteoIQ. Our analysis indicates deviation from expected log ratios appear to stem from the lower intensity measurements. This potentially limits utilization of this quantification approach in analysis of proteins at low copy number or sample limited acquisitions.