Finding patterns in unassigned peak lists of protein and peptide fragment mass spectra using frequent itemset mining techniques

摘要

Analysis of unassigned peaks in large numbers of peptide mass fingerprint spectra reveals several highly frequent masses. Frequent item set mining techniques yields a large number of peaks that frequently co-occur in unassigned peak lists of individual MS spectra. Prioritization, redundancy removal and assesment of statistical significance are remaining challenges in mass spectrometry-based frequent item set mining. Several more advanced techniques exist that need to be evaluated. Further evaluation of retrieved patterns will be based on MSMS analyses. We need approaches to rapidly visualise, browse, filter and investigate retrieved patterns.