Proteomic mass spectra classification using decision tree based ensemble methods
(2005)
Journal - Bioinformatics
Abstract :
Motivation: Modern mass spectrometry allows the determinationof proteomic fingerprints of body fluids like serum, salivaor urine. These measurements can be used in many medical applicationsin order to diagnose the current state or predict the evolutionof a disease. Recent developments in machine learning allowone to exploit such datasets, characterized by small numbersof very high-dimensional samples.Results: We propose a systematic approach based on decisiontree ensemble methods, which is used to automatically determineproteomic biomarkers and predictive models. The approach isvalidated on two datasets of surface-enhanced laser desorption/ionizationtime of flight measurements, for the diagnosis of rheumatoidarthritis and inflammatory bowel diseases. The results suggestthat the methodology can handle a broad class of similar problems.Supplementary information: Additional tables, appendicies anddatasets may be found at http://www.montefiore.ulg.ac.be/~geurts/Papers/Proteomic-suppl.htmlContact:p.geurts{at}ulg.ac.be