s'authentifier
version française rss feed
HAL : hal-00559580, version 2

Fiche détaillée  Récupérer au format
PLoS ONE (2011) 6(12): e28210
Versions disponibles :
The influence of feature selection methods on accuracy, stability and interpretability of molecular signatures
Anne-Claire Haury ( ) 1, 2, Pierre Gestraud 2, Jean-Philippe Vert 1, 2
(21/12/2011)

Motivation: Biomarker discovery from high-dimensional data is a crucial problem with enormous applications in biology and medicine. It is also extremely challenging from a statistical viewpoint, but surprisingly few studies have investigated the relative strengths and weaknesses of the plethora of existing feature selection methods. Methods: We compare 32 feature selection methods on 4 public gene expression datasets for breast cancer prognosis, in terms of predictive performance, stability and functional interpretability of the signatures they produce. Results: We observe that the feature selection method has a significant influence on the accuracy, stability and interpretability of signatures. Simple filter methods generally outperform more complex embedded or wrapper methods, and ensemble feature selection has generally no positive effect. Overall a simple Student's t-test seems to provide the best results. Availability: Code and data are publicly available at http://cbio.ensmp.fr/~ahaury/.
1 :  Centre de Bioinformatique (CBIO)
MINES ParisTech - École nationale supérieure des mines de Paris
2 :  Cancer et génôme: Bioinformatique, biostatistiques et épidémiologie d'un système complexe
INSERM : U900 – Institut Curie – MINES ParisTech - École nationale supérieure des mines de Paris
Sciences du Vivant/Bio-Informatique, Biologie Systémique

Informatique/Bio-informatique

Statistiques/Machine Learning

Statistiques/Applications
biomarker discovery – gene expression – microarrays – feature selection – signature – breast cancer
Liste des fichiers attachés à ce document : 
ANNEX
supplementary.pdf(1.5 MB)
PDF
techreport2010.pdf(250.2 KB)
PS
techreport2010.ps(1.4 MB)