A model of random sequences for de novo peptide sequencing
peer-reviewed · Third IEEE Symposium on Bioinformatics and Bioengineering, 2003. Proceedings · 2003
| Date | 2003-11-21 |
| Type | peer-reviewed |
| Venue | Third IEEE Symposium on Bioinformatics and Bioengineering, 2003. Proceedings |
| Publisher | IEEE Comput. Soc |
| Contribution | algorithm |
| DOI | 10.1109/bibe.2003.1188948 |
| Citations (OpenAlex) | 7 |
Abstract
We present a model for the probability of random sequences appearing in product ion spectra obtained from tandem mass spectrometry experiments using collision-induced dissociation. We demonstrate the use of these probabilities for ranking candidate peptide sequences obtained using a de novo algorithm. Sequence candidates are obtained from a spectrum graph that is greatly reduced in size from those in previous graph-theoretical de novo approaches. Evidence of multiple instances of subsequences of each candidate, due to different fragment ion type series as well as isotopic peaks, is incorporated in a hierarchical scoring scheme. This approach is shown to be useful for confirming results from database search and as a first step towards a statistically rigorous de novo algorithm.
Methods and tools
- Hierarchical subsequence de novo scoring (PNNL): PNNL de novo approach that finds sequences and subsequences hierarchically in a reduced spectrum graph, using multiple ion series and natural isotope peaks, and ranks candidates by the probability of random matches.