Premium
Clustering Gene Expression Data using a Posterior Split‐Merge‐Birth Procedure
Author(s) -
SARAIVA ERLANDSON F.,
MILAN LUIS A.
Publication year - 2012
Publication title -
scandinavian journal of statistics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.359
H-Index - 65
eISSN - 1467-9469
pISSN - 0303-6898
DOI - 10.1111/j.1467-9469.2011.00765.x
Subject(s) - merge (version control) , cluster analysis , expression (computer science) , posterior probability , data mining , mathematics , computer science , computational biology , statistics , bayesian probability , biology , information retrieval , programming language
. DNA array technology is an important tool for genomic research due to its capa‐city of measuring simultaneously the expression levels of a great number of genes or fragments of genes in different experimental conditions. An important point in gene expression data analysis is to identify clusters of genes which present similar expression levels. We propose a new procedure for estimating the mixture model for clustering of gene expression data. The proposed method is a posterior split‐merge‐birth MCMC procedure which does not require the specification of the number of components, since it is estimated jointly with component parameters. The strategy for splitting is based on data and on posterior distribution from the previously allocated observations. This procedure defines a quick split proposal in contrary to other split procedures, which require substantial computational effort. The performance of the method is verified using real and simulated datasets.