Premium
Projection‐based informatics approaches to serum/plasma metabolomics data: applications to biomarkers for caloric intake in rats
Author(s) -
Kristal Bruce S,
Shurubor Yevgeniya,
Marur Vasant
Publication year - 2007
Publication title -
the faseb journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.709
H-Index - 277
eISSN - 1530-6860
pISSN - 0892-6638
DOI - 10.1096/fasebj.21.5.a310
Subject(s) - projection (relational algebra) , metabolomics , partial least squares regression , linear discriminant analysis , caloric theory , nutritional epidemiology , statistics , class (philosophy) , computer science , artificial intelligence , mathematics , bioinformatics , medicine , biology , epidemiology , algorithm
Dietary or caloric restriction (DR) is the most potent and reproducible known means of reducing cancer risk in mammals. We have developed serum metabolomic profiles that can identify ad libitum fed and caloric‐restricted rats with a high degree of accuracy. These profiles are being adapted for use in human epidemiology studies, given the increased risk of disease associated with excess weight. Partial Least Squares Projection to Latent Structures Discriminant Analysis (PLS‐DA), a projection method optimized for class separation built models with >95% accuracy in distinguishing groups without obvious cohort interference, but these models are potentially complex to apply when testing models generated on discrete classes on datasets with more intermediate values. We will describe comparative analysis of PLS‐DA with O‐PLS (Orthogonal‐PLS) to determine the utility of this latter approach for this class of problem.