An artificial neural network is developed for predicting which fragment is charged and which fragment is neutral for lipid fragment pairs produced from a liquid chromatography tandem mass spectrometry simulation process. This charge predictor is integrated into software developed at PNNL for in silico spectra generation and identification of metabolites known as Met ISIS. To test the effect of including charge prediction in Met ISIS, 46 lipids are used which show a reduction in false positive identifications when the charge predictor is utilized.
Revised: July 24, 2014 |
Published: December 18, 2011
Citation
Schrom B.T., L.J. Kangas, B. Ginovska, T.O. Metz, and J.H. Miller. 2011.Charge Prediction of Lipid Fragments in Mass Spectrometry. In 10th International Conference on Machine Learning and Applications and Workshops (ICMLA 2011), December 18-21, 2011, Honolulu, Hawaii, 2, 186-188. Piscataway, New Jersey:IEEE.PNNL-SA-79072.doi:10.1109/ICMLA.2011.45