Constrained De Novo Sequencing of Peptides with Application to Conotoxins

Abstract

We describe algorithms for incorporating prior sequence knowledge into the candidate generation stage of de novo peptide sequencing by tandem mass spectrometry. We focus on two types of prior knowledge: homology to known sequences encoded by a regular expression or position-specific score matrix, and amino acid content encoded by a multiset of required residues. We show an application to de novo sequencing of cone snail toxins, which are molecules of special interest as pharmaceutical leads and as probes to study ion channels. Cone snail toxins usually contain 2, 4, 6, or 8 cysteine residues, and the number of residues can be determined by a relatively simple mass spectrometry experiment. We show here that the prior knowledge of the number of cysteines in a precursor ion is highly advantageous for de novo sequencing.