Sequence Database Compression for Peptide Identification from Tandem Mass Spectra

The identification of peptides from tandem mass spectra is an important part of many high-throughput proteomics pipelines. In the high-throughput setting, the spectra are typically identified using software that matches tandem mass spectra with putative peptides from amino-acid sequence databases. The effectiveness of these search engines depends heavily on the completeness of the amino-acid sequence database used, but suitably complete amino-acid sequence databases are large, and the sequence database search engines typically have search times that are proportional to the size of the sequence database. We demonstrate that the peptide content of an amino-acid sequence database can be represented by a reformulated amino-acid sequence database containing fewer amino-acid symbols than the original. In some cases, where the original amino-acid sequence database contains many redundant peptides, we have been able to reduce the size of the amino- acid sequence to almost half of its original size. We develop a lower bound for achievable compression and demonstrate empirically that regardless of the peptide redundancy of the original amino-acid sequence database, we can compress the sequence to within 15-25% of this lower bound. We believe this may provide a principled way to combine amino-acid sequence data from many sources without unduly bloating the resulting sequence database with redundant peptide sequences.