Differential proteomics via probabilistic peptide identification scores.

Relative quantitation is key to enable differential proteomics and hence answer biological questions by comparing samples. Classical approaches involve stable isotope labeling with/without spiked standards. Although stable isotopes may lead to precise results, their application is not straightforward. In Proteomics, 2004, 4, 2333-2351, we proposed an approach where we summed peptide identification scores to derive a semiquantitative abundance indicator. In this study, we combine such an indicator with a statistical test to detect differentially expressed proteins. We demonstrate the effectiveness of this method by using mixtures of purified proteins and human plasma spiked with proteins at low-nanomolar concentrations. The impact of the number of repeated experiments is discussed, and we show that the statistical test we use performs well with two to three repetitions, whereas a classical t-test would require at least four repetitions to achieve the same performance. Typically, 2.5-5-fold changes are detected with 90-95% confidence in human plasma. The method is finally characterized by deriving estimates of its false positive and negative rates. This new characterization is valid for a wider class of methods such as spectrum sampling (Liu, H.; Sadygov, R. G.; Yates, J. R. III. Anal. Chem. 2004, 76, 4193-4201).