Mathematical Modeling and Analysis of Mass Spectrometry Data in Workflows for the Discovery of Biomarkets in Breast Cancer

摘要

The major achievement for the last year was the development of the workflow for the LC-MS/MS proteomic experiment and testing it on the real data. This included setting up an appropriate mathematical/statistical scheme of random-effect models. The overall significance of multiple hypothesis testing was controlled by the false discovery rate approach. To improve the statistical power and clinical interpretation of the results, the protein identifications were mapped to high quality publicly available data sources, and differential quantification was rolled up to the gene level. The technique similar to the gene set enrichment analysis in microarrays is used to find statistically significant differences between the protein families. To achieve a dataset with minimal false identifications, functional annotations from the GeneOntology and the HUPO project were added as well as known plasma concentration from the literature. To understand the biological relevance of the differentially expressed proteins we used several functional annotation tools. We plan to complement the mass-spectrometry data with metabolomic analysis, and will provide additional data from protein microarrays.e.