Linked Life Data

19634197

Source:http://linkedlifedata.com/resource/pubmed/id/19634197

Statements in which the resource exists as a subject.
PredicateObject
rdf:type
lifeskim:mentions
pubmed:issue
4
pubmed:dateCreated
2009-7-27
pubmed:abstractText
DNA microarrays have become a powerful tool to describe gene expression profiles associated with different cellular states, various phenotypes and responses to drugs and other extra- or intra-cellular perturbations. In order to cluster co-expressed genes and/or to construct regulatory networks, definition of distance or similarity between measured gene expression data is usually required, the most common choices being Pearson's and Spearman's correlations. Here, we evaluate these two methods and also compare them with a third one, namely Hoeffding's D measure, which is used to infer nonlinear and non-monotonic associations, i.e. independence in a general sense. By comparing three different variable association approaches, namely Pearson's correlation, Spearman's correlation and Hoeffding's D measure, we aimed at assessing the most appropriate one for each purpose. Using simulations, we demonstrate that the Hoeffding's D measure outperforms Pearson's and Spearman's approaches in identifying nonlinear associations. Our results demonstrate that Hoeffding's D measure is less sensitive to outliers and is a more powerful tool to identify nonlinear and non-monotonic associations. We have also applied Hoeffding's D measure in order to identify new putative genes associated with tp53. Therefore, we propose the Hoeffding's D measure to identify nonlinear associations between gene expression profiles.
pubmed:language
eng
pubmed:journal
pubmed:citationSubset
IM
pubmed:chemical
pubmed:status
MEDLINE
pubmed:month
Aug
pubmed:issn
0219-7200
pubmed:author
pubmed:issnType
Print
pubmed:volume
7
pubmed:owner
NLM
pubmed:authorsComplete
Y
pubmed:pagination
663-84
pubmed:meshHeading
pubmed:year
2009
pubmed:articleTitle
Comparing Pearson, Spearman and Hoeffding's D measure for gene expression association analysis.
pubmed:affiliation
Human Genome Center, University of Tokyo, 4-6-1 Shirokanedai, Minato-ku, Tokyo, 108-8639, Japan. afujita@ims.u-tokyo.ac.jp
pubmed:publicationType
Journal Article, Comparative Study, Research Support, Non-U.S. Gov't, Evaluation Studies