|
Activity Number:
|
342
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #307062 |
|
Title:
|
Gene Selection Using a Modified Mahalanobis Distance
|
|
Author(s):
|
Xiwen Ma*+
|
|
Companies:
|
University of Wisconsin-Madison
|
|
Address:
|
1300 University Ave., Madison, WI, 53706,
|
|
Keywords:
|
Mahalanobis distance ; eigenvalue, eigenvector ; maximum likelihood estimate ; non-parametric
|
|
Abstract:
|
A fundamentally important question in microarray experiments is to locate differentially expressed genes for disease diagnosis. We consider the experiment which compares two sources of RNA between diseased and healthy subjects. The key problem is to well define a criterion of discrimination and to detect an optimal subset of genes according to it. We propose a modified Mahalanobis distance which is independent of distribution assumptions and can be used in detecting an optimal subset of genes. We also provide a stepwise method to locate the optimum subset and apply to well known benchmark public datasets and compare with existing approaches.
|
