|Home | About | Journals | Submit | Contact Us | Français|
Methods based on isotope labeling and LC-MS/MS analysis are becoming a standard approach for the simultaneous identification and quantification of proteins on the proteomic scale, as they provide high accuracy, precision, sensitivity, speed, and compatibility with protein pre-fractionation. However, due to the high complexity of these experiments, a software platform is required that allows for efficient data acquisition, interpretation, and validation.
We describe such a validation platform for quantitative proteomics, which supports different types of mass spectrometers (e.g., MALDI-TOF/TOF, ESI-ion trap, -Q-OTOF, and -FTMS) and any label chemistry including multiplex labels. The software assists in the optimization of LC and MS instrument-specific parameters by providing 2D views on the whole LC-MS/MS dataset (SurveyViewer). The SurveyViewer facilitates tasks such as the fast assessment of the mass calibration or the fragmentation efficiency of isobaric labeling tags.
Protein regulations are derived from peptide regulation distributions using lognormal theory and median statistics. Median statistics is used in addition for the automated elimination of outliers. Box and whisker plots and bar graphs assist in evaluating the quantitative results and providing the basis for the manual acceptance or rejection of individual quantitative data.
Direct links to sequence-annotated raw spectra of a particular regulated protein or peptide with plots of the identified amino acid sequence allow for efficient case study and curation of unexpected results, and the interactive validation of uncertain results.
The described features of the software platform WARP-LC are exemplified in LC-MS/MS studies of protein mixtures using ICPL, iTRAQ, and SILAC. Essential methodological issues such as quantification based on median vs. average statistics and peak areas vs. peak intensities were evaluated. It appears that the monoisotopic peak intensities of de-isotoped isotopic clusters are preferable for quantification, as the average quantification errors are approximately 20% lower.