Abstract
Current proteomics workflows are very diverse in terms of experimental approaches, methods of extracting quantitative information, and how this quantitative information is utilized in the experiment. This presentation will start with a brief overview of quantitative proteomics, with a focus on label-free strategies and accurate computational analysis of the data. We will discuss both continuous (e.g. peptide intensity-based) and count-based (spectral counting) quantitative strategies. In comparing these two approaches, we will focus on practical aspects of quantitative proteomics, such as most critical data analysis steps, computational strategies and tools that are available, and how various factors (e.g. mass accuracy) affect the quality of quantitative data. We will also discuss specific challenges of quantitative proteomic experiments, including: 1) The problem of small number of replicates, and how it affects the accuracy of quantitation; 2) How to use quantitative information from shared peptides, i.e. peptides whose sequence is present in multiple proteins. 3) How to determine where to draw the threshold for calling a protein to be enriched/differentially expressed 4) Methods to estimate false discovery rates in the context of quantitation. To address these challenges, we developed a suite of freely available computational tools. These tools include ABACUS and QSPEC for processing and statistically modeling spectral count data, and more recently QPROT for continuous data. QPROT extends the Bayesian hierarchical model of QSPEC for applications to precursor ion intensity data as well as continuously normalized count data, while accounting for the missing data properly. We demonstrate QPROT and perform comparisons with other methods using data from a recent clinical proteomic technology assessment for cancer (CPTAC) study. We will also present the results of comparative analysis of different label-free quantitation methods, and comparative analysis of absolute protein and mRNA data, using publicly available proteomic and transcriptomic data on mouse mitochondria.