Protein De novo Sequencing

Abstract

In the proteomic mass spectrometry field, peptide and protein identification can be classified into two categories: database search that relies on existing peptide and protein databases and de novo sequencing with no prior knowledge. There are many unknown protein sequences in nature, especially those proteins that play an vital role in drug development pipelines, such as monoclonal antibodies and venoms. To sequence these unknown proteins, de novo sequencing is a necessity. There have been standard algorithms for de novo sequencing a short peptide from its tandem mass spectrum (MS/MS). However, the de novo sequencing of a whole protein is still in its infancy. The most promising method is to digest the protein into overlapping short peptides with different enzymes. After each peptide is de novo sequenced with MS/MS, these overlapping peptides are then assembled together either manually or with a computer algorithm. Such an automated assembly algorithm becomes the main purpose of this thesis. Compared to the DNA sequence assembly counterpart, the main challenges are the high error rates and the short sequence length of each de novo peptide. To meet these challenges, novel scoring methods and algorithms are proposed and a software program is developed. The program is tested on a standard data set and demonstrates superior performance when compared to the state-of-the-art.