CFM-ID applied to CASMI 2014
- Felicity Allen
- Russ Greiner, Dept of Computing Science; PI of AICML
- David S. Wishart, Departments of Computing Science and Biology, University of Alberta
Background:: The CASMI competition is an annual international competition in which contestants must provide putative chemical structures for each of a set of challenge mass spectra.
Methods: For each target spectrum, we used CFM-ID to provide a score for each candidate compound, based on its ESI-MS/MS spectrum. We combined this with two other scoring components that made use of additional meta information provided with each challenge. We also investigated the role of this meta information in the results obtained.
Results: The resulting method correctly identified 24 out of 42 challenge compounds, and ranked the correct candidate in the top 10 in 33 challenges. The use of the meta information was found to play a significant role in the identification performance.
Conclusion: This was sufficient to win the CASMI competition, which is evolving to become an important arena for assessment of progress in this field.
Citation
F. Allen, R. Greiner, D. Wishart. "CFM-ID applied to CASMI 2014". Current Metabolomics, 5(1), pp 35-39, June 2016.| Keywords: | Mass spectrometry, CASMI, small molecule identification, machine learning, probabilistic models, metabolomics, bioinformatics |
| Category: | In Journal |
| Web Links: | DOI |
| Journal |
BibTeX
@article{Allen+al:CurrentMeta16,
author = {Felicity Allen and Russ Greiner and David S. Wishart},
title = {CFM-ID applied to CASMI 2014},
Volume = "5",
Number = "1",
Pages = {35-39},
journal = {Current Metabolomics},
year = 2016,
}Last Updated: February 10, 2020Submitted by Sabina P
