Bioinformatics Advance Access published online on June 5, 2008
Bioinformatics, doi:10.1093/bioinformatics/btn245
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Knowledge - based Gene Expression Classification via Matrix Factorization
a CIML/Biophysics, University of Regensburg, D-93040 Regensburg, Germany, b CMB/IBI, GSF Munich, Germany, c IEETA/DETI, Universidade de Aveiro, 3810-193 Aveiro, Portugal, d G. Schmitz, Clinical Chemistry, University Hospital Regensburg, D-93042 Regensburg, Germany, e M. Stetter, Siemens Corporate Technology, Siemens AG, Munich, f P. Gómez Vilda, DATSI/FI, Universidad Politécnica de Madrid, E-18500 Madrid, Spain
To whom correspondence should be addressed. Prof. Elmar Lang, E-mail: elmar.lang{at}biologie.uni-regensburg.de
 |
Abstract |
|---|
Motivation: Modern machine learning methods based on matrix decomposition techniques, like Independent Component Analysis (ICA) or Nonnegative Matrix Factorization (NMF), provide new and efficient analysis tools which are currently explored to analyze gene expression profiles (Quackenbush, 2001). These exploratory feature extraction techniques yield expression modes (ICA) or metagenes (NMF). These extracted features are considered indicative of underlying regulatory processes. They can as well be applied to the classification of gene expression data sets by grouping samples into different categories for diagnostic purposes or group genes into functional categories for further investigation of related metabolic pathways and regulatory networks.
Results: In this study we focus on unsupervised matrix factorization techniques and apply ICA and sparse NMF to microarray data sets. The latter monitor the gene expression levels of human peripheral blood cells during differentiation from monocytes to macrophages. We show that these tools are able to identify relevant signatures in the deduced component matrices and extract informative sets of marker genes from these gene expression profiles. The methods rely on the joint discriminative power of a set of marker genes rather than on single marker genes. With these sets of marker genes, corroborated by leave-one-out or random forest cross - validation, the datasets could easily be classified into related diagnostic categories. The latter correspond to either monocytes vs macrophages or healthy vs Niemann Pick C disease patients.
Contact: elmar.lang{at}biologie.uni-regensburg.de
Associate Editor: Dr. Olga Troyanskaya
Received on September 24, 2008; revised on May 14, 2008; accepted on May 23, 2008
CiteULike 
Connotea 
Del.icio.us What's this?