MS/MS Spectra Interpretation as a Statistical–Mechanics Problem
peer-reviewed · Analytical Chemistry · 2013
| Date | 2013-05-21 |
| Type | peer-reviewed |
| Venue | Analytical Chemistry |
| Publisher | American Chemical Society (ACS) |
| Contribution | algorithm |
| DOI | 10.1021/ac4005666 |
| Citations (OpenAlex) | 4 |
| Venue 2-year citedness | 7.10 |
Abstract
We describe a new method for peptide sequencing based on the mapping of the interpretation of tandem mass spectra onto the analysis of the equilibrium distribution of a suitably defined physical model, whose variables describe the positions of the fragmentation sites along a discrete mass index. The model is governed by a potential energy function that, at present, we derive ad hoc from the distribution of peaks in a data set of experimental spectra. The statistical-physics perspective prompts for a consistent and unified approach to de novo and database-search methods, which is a distinctive feature of this approach over alternative ones: the characterization of the ground state of the model allows the de novo identification of the precursor peptide; the study of the thermodynamic variables as a function of the (fictitious) temperature gives insight on the quality of the prediction, while the probability profiles at nonzero temperature reveal, on one hand, which fragments are more reliably predicted. On the other hand, they can be used as a spectrum-adapted, a posteriori score for database search. Results obtained with two different test data sets reveal a performance similar to that of other de novo and database-search methods, which is reasonable, given the lack of an aggressive optimization of the energy function at this stage. An important feature of the method is that it is quite general and can be applied with different choices of the energy function: we discuss its possible improvements and generalizations.
Methods and tools
- Statistical-mechanics MS/MS interpretation: Treats fragmentation sites as a physical system; the ground state yields the de novo peptide and thermodynamic variables estimate confidence.