Statistical Modeling of Protein Modification Regulation

The understanding of regulatory mechanisms of biological processes is at the heart of many research efforts in medicine and molecular biology. One important means of regulating biochemical reactions taking place in living cells is through the control of protein posttranscriptional modifications (PTM), which activate, deactivate, or change the fate of proteins in many ways. The changes induced by PTMs result in the adjustment or deregulation of metabolic or signaling pathways, the most prevalent example of which is the action of phosphorylation in signaling and its impact in cancer. Proteomics technologies provide researchers with powerful tools to obtain quantitative measures of protein modification variations under different conditions, e.g. healthy versus diseased patients. Enormous flows of data are generated by mass spectrometry and they are difficult to analyze without proper statistical knowledge but rather little is known about the specific statistical distributions required to describe them. Among the various quantitative proteomics platforms, multiplexed technologies such as iTRAQ or TMT are widely spread and very appropriate to compare biological and medical samples in multiple conditions. Therefore, we decided to take iTRAQ/TMT as a model and to assemble complete statistical models of PTM regulation as well as bioinformatics tools intended to facilitate the extraction of meaningful knowledge from such huge datasets. We have already published fundamental models for the analysis of protein regulation (Breitwieser et al., J Proteome Res, 2011) we will work with similar techniques to capture significant changes at the peptide modification site-level. We will also develop new statistical models to describe multiple iTRAQ/TMT experiments and hence compare larger collections of samples in a rigorous manner. Specific test datasets will be generated to support the development of the project. In Breitwieser et al. we showed that the application of sound statistical methods as well as the careful modeling of technical and biological sources of variability resulted in a highly sensitive method able to control its false detection rate. We postulate that the same approach will yield useful new useful statistical insights PTM regulation data. The project also addresses the need of implementation such that other researchers can benefit from our work. We will continue releasing open source Bioconductor packages for the R platform as we did previously (Breitwieser et al.). R is not only one of the prevalent bioinformatics platform but it attracts always more biologists who want to analyze data by themselves and are supported by well designed packages that leverage the technical difficulty. Finally, we will provide scripts to automate data processing.

Proteomics-Technologien bieten Forschern leistungsfähige Werkzeuge zum quantitativen Messen der Veränderung der Protein-Modifikation in verschiedenen Bedingungen, z. B. in gesunden gegenüber kranken Patienten. Enorme Datenflüsse werden durch Massenspektrometrie erzeugt, und können die ohne richtigen statistischen Methoden und Werkzeuge der Bioinformatik nicht analysiert werden Allerdings ist nur wenig über die statistischen Verteilungen bekannt welche die Daten beschreiben können. Für den Vergleich werden Peptide oder Proteine unterschiedlicher Proben gekennzeichnet. Die Multiplex-Technologien iTRAQ und TMT sind weit verbreitet und gut geeignet, um biologische und medizinische Proben in mehrere Bedingungen zu vergleichen. Wir konzentrieren uns auf die Datenanalyse dieser chemischen isobarischen Peptidmarkierungsmethoden um vollständige statistische Modelle der PTM Regulierung sowie Bioinformatik-Tools zur Gewinnung von aussagekräftigem Wissen aus solchen riesigen Datenmengen zu generieren.