Summary
Accurate data processing and statistical analysis of label-free and stable isotope labeled LC-MS data used in comparative profiling is of primary importance for proper interpretation of proteomics experiences and to suitably answer the underlying biological questions. In addition to the need of the development of specific algorithms solving particular problems related to the data properties, the performance of data processing algorithms and statistical analysis have to be evaluated in a systematic manner.
The project proposal intends to develop modularization of existing open-source data processing pipelines within a framework enabling to interconnect those modules combined with a mechanism to select the best combination of data processing modules for given datasets using spiking experiences.
Development of time alignment algorithms enabling proper analysis of charge state and isotopic distribution shifted LC-MS data and dataset obtained from other laboratories with different sample preparation techniques having a low fraction of compound overlap, will enable to expand accurate quantitative proteomics for a wider range of proteomics applications and will increase knowledge extraction for life science research. This will be supported by developing data integration techniques for data obtained from multidimensional techniques. Finally specific bioinformatics tools for quality control of data processing steps to increase the measured dynamic concentration range of quantitative proteomics techniques will be developed.
Statistical simulation assisted through spiked datasets will enable us to asses the global and individual performance of data processing, statistical analysis and validation modules. Statistical simulation using analytical and biological variance, measured dynamic concentration range and a noise model obtained from real datasets will enable us to select the best statistical strategy to detect class compound concentration differences.