Enhanced protein identification through automated search parameter selection

Abstract

Mass spectrometry-based proteomics plays a critical role in identifying and quantifying proteins. Proteomics search engines, integral to this process, require meticulous parameter selection to achieve accurate results. However, the large number of available search parameters makes it challenging to manually choose the optimal combinations. This thesis focuses on optimizing and automating the parameter selection process. The main idea is to select and search a subset of the data that preserves the properties of the complete dataset, enabling efficient parameter exploration while minimizing both computational resources and time requirements. The approach has been validated using various mass spectrometry datasets from PRIDE analyzed via the SearchGUI and PeptideShaker framework. Easy support for testing multiple parameter combinations ultimately leads to overall better parameter selection, thus enhancing protein identification accuracy and workflow efficiency, which in turn contributes to getting the most out of valuable biological samples.