PepTiger: Search Engine for Error-Tolerant Protein Identification from de Novo Sequences

Abstract

In recent years a number of de novo sequencing software products became available providing possible partial or complete amino acid sequence tags for MS/MS spectra of peptides. However, for a variety of reasons including spectral chemical noise and imperfect fragmentation these sequence tags almost always contain errors. Additional difficulties arise from actual protein sequence variation and post-translational modifications. We present a search engine named PepTiger which is capable of correctly matching de novo sequence tags with errors to protein sequences in a protein database. The algorithm is based on approximate string matching followed by a novel scoring procedure which takes into account mass differences and the string distance between de novo sequence and matched peptides and similarities between theoretical and experimental MS/MS spectra. Comparison of PepTiger with other protein identification software shows that PepTiger is better able to assign de novo sequence tags with errors to the correct peptide sequences.