SlimPLS A Method for Feature Selection in Gene Expression-Based Disease Classification 英文参考文献.docVIP

SlimPLS:AMethodforFeatureSelectioninGene Expression-BasedDiseaseClassification MichaelGutkin1,RonShamir1*,GideonDror2 1BlavatnikschoolofComputerScience,TelAvivUniversity,TelAviv,Israel,2SchoolofComputerScience,TheAcademicCollegeofTel-Aviv-Yaffo,Tel-Aviv,Israel Abstract A major challenge in biomedical studies in recent years has been the classification of gene expression profiles into categories,suchascasesandcontrols.Thisisdonebyfirsttrainingaclassifierbyusingalabeledtrainingsetcontaining labeled samples from the two populations, and then using that classifier to predict the labels of new samples. Such predictionshaverecentlybeenshowntoimprovethediagnosisandtreatmentselectionpracticesforseveraldiseases.This procedure is complicated, however, by the high dimensionality if the data. While microarrays can measure the levels of thousands of genes per sample, case-control microarray studies usually involve no more than several dozen samples. Standardclassifiersdonotworkwellinthesesituationswherethenumberoffeatures(geneexpressionlevelsmeasuredin thesemicroarrays)farexceedsthenumberofsamples.Selectingonlythefeaturesthataremostrelevantfordiscriminating betweenthetwocategoriescanhelpconstructbetterclassifiers,intermsofbothaccuracyandefficiency.Inthisworkwe developedanovelmethodformultivariatefeatureselectionbasedonthePartialLeastSquaresalgorithm.Wecomparedthe method’s variants with common feature selection techniques across a large number of real case-control datasets, using severalclassifiers.Wedemonstratetheadvantagesofthemethodandthepreferablecombinationsofclassifierandfeature selectiontechnique. Citation: Gutkin M, Shamir R, Dror G (2009) SlimPLS: AMethodfor Feature Selection inGeneExpression-Based Disease Classification. PLoS ONE 4(7): e6416. doi:10.1371/journal.pone.0006416 Editor:MagnusRattray,UniversityofManchester,UnitedKingdom ReceivedFebruary23,2009;AcceptedJune11,2009;PublishedJuly29,2009 Copyright: ? 2009 Gutkin et al. This is an open-access article distribute