Premium
Gene expression profile analysis using discrimination and fuzzy classification methods
Author(s) -
Kapil Ankur,
Gudi Ravindra D.,
Noronha Santosh B.
Publication year - 2006
Publication title -
asia‐pacific journal of chemical engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.348
H-Index - 35
eISSN - 1932-2143
pISSN - 1932-2135
DOI - 10.1002/apj.12
Subject(s) - cluster analysis , principal component analysis , fuzzy logic , identification (biology) , linear discriminant analysis , computer science , data mining , expression (computer science) , curse of dimensionality , artificial intelligence , pattern recognition (psychology) , projection (relational algebra) , machine learning , biology , algorithm , botany , programming language
There is a huge incentive for gene expression analysis and identification of biologically meaningful clusters from microarray data. However, the high dimensionality of the data poses challenges for this task. Here, to reduce this problem of irrelevant dimensions, we consider three different projection methods, viz. principal components analysis (PCA), correspondence analysis (CA), and multiple discriminant analysis (DA). To account for the possibility of pleiotropy, where the expression of certain genes may be related to more than one phenotypical condition, we use fuzzy clustering on the lower dimensional space generated by PCA, CA, and DA. Fuzzy clustering permits partial belonging of an attribute, such as gene expression, to different functionalities and hence is eminently suited for this task. To determine the optimum number of clusters, we evaluate various cluster validity indices. In this paper, we compare these methodologies when applied to the data generated by a genetic network simulator (eXPatGen) and also to the experimental micro array data available for yeast S. cerevisiae . © 2006 Curtin University of Technology and John Wiley & Sons, Ltd.