Detecting reliable gene interactions by a hierarchy of Bayesian network classifiers

Abstract

The main purpose of a gene interaction network is to map the relationships of the genes that are out of sight when a genomic study is tackled. DNA microarrays allow the measure of gene expression of thousands of genes at the same time. These data constitute the numeric seed for the induction of the gene networks. In this paper, we propose a new approach to build gene networks by means of Bayesian classifiers, variable selection and bootstrap resampling. The interactions induced by the Bayesian classifiers are based both on the expression levels Keywords:an d on the phenotype information of the supervised variable. Feature selection and bootBayesian network classifiers strap resampling add reliability and robustness to the overall process removing the false Robust arc identification positive findings. The consensus among all the induced models produces a hierarchy of Gene interactions dependences and, thus, of variables. Biologists can define the depth level of the model hierar-DNA microarrays chy so the set of interactions and genes involved can vary from a sparse to a dense set. ExperKnowledge discovery imental results show how these networks perform well on classification tasks. The biological validation matches previous biological findings and opens new hypothesis for future studies.

A Study of Network-based Approach for Cancer Classification

“…To prevent bias in proposed model, [37] never used a priori biological information since the available biological networks not fully completed [46]. Thus, phenotype distribution will be used for such purpose similar to what have ever done by [47].…”

“…Actually, not many researchers focusing on the statistical information that the comparison of different sample types contributes rather than just look for differently expressed genes to build their model [37]. Because of it, the methods that can deal with this matter is desired not only for precise accuracy but computation time as well.…”

“…Emphasized again, network-based here means that interactions among genes is taken into account [35][36][37] for informative gene selection in term of building classifier. It is because disease phenotype cannot be determined by only single gene [20], but need at least interaction information from a pair of genes.…”

A Bayesian network approach for modeling local failure in lung cancer

“…van Gerven et al (2008) demonstrated the development of a prognostic model for carcinoid patients using dynamic Bayesian networks. Arma ñ anzas et al (2008) used a hierarchical Bayesian structure learning method to detect gene interactions. Smith et al (2009) developed a prognostic model for prostate cancer with intensity modulated radiation therapy (IMRT) plans and calculated a quality-adjusted life expectancy for each plan using Bayesian networks.…”

Bayesian networks classifiers for gene-expression data

“…Armañanzas et al [27] use a KDB classifier, with k = 4. Bootstrap resampling with 1000 iterations is used to add reliability and robustness to the overall process.…”