Automated procedure for candidate compound selection in GC-MS metabolomics based on prediction of Kovats retention index

• DOI: 10.1093/bioinformatics/btp056
• Publication Date: 2009-01-29T01:31:13Z
• Authors: V. V. Mihaleva; H. A. Verhoeven; R. C. H. de Vos; R. D. Hall; R. C. H. J. van Ham
• Keywords: Kovats retention index; Data set

Abstract Motivation: Matching both the retention index (RI) and mass spectrum of an unknown compound against a spectral reference library provides strong evidence for correct identification that compound. Data on indices are, however, available only small fraction compounds in such libraries. We propose quantitative structure-RI model enables ranking filtering putative identifications which predicted RI falls outside predefined window. Results: constructed multiple linear regression support vector (SVR) models using set descriptors obtained with genetic algorithm as variable selection method. The SVR is significant improvement over previous built structurally diverse it covers large range (360–4100) values gives better prediction isomer compounds. hit list reduction varied from 41% to 60% depended size original list. Large lists were reduced greater extend compared lists. Availability: http://appliedbioinformatics.wur.nl/GC-MS Contact: [e-mail] Supplementary information: data are at Bioinformatics online.