Indexing and Searching a Mass Spectrometry Database

Database preprocessing in order to create an index often permits considerable speedup in search compared to the iterated query of an unprocessed database. In this paper we apply index-based database lookup to a range search problem that arises in mass spectrometry-based proteomics: given a large collection of sparse integer sets and a sparse query set, find all the sets from the collection that have at least k integers in common with the query set. This problem arises when searching for a mass spectrum in a database of theoretical mass spectra using the shared peaks count as similarity measure. The algorithms can easily be modified to use the more advanced shared peaks intensity measure instead of the shared peaks count. We introduce three different algorithms solving these problems. We conclude by presenting some experiments using the algorithms on realistic data showing the advantages and disadvantages of the algorithms.