A Robust Multifactor Dimensionality Reduction Method for Detecting Gene-Gene Interactions with Application to the Genetic Analysis of Bladder Cancer Susceptibility

Abstract

SummaryA central goal of human genetics is to identify susceptibility genes for common human diseases. An important challenge is modelling gene-gene interaction or epistasis that can result in nonadditivity of genetic effects. The multifactor dimensionality reduction (MDR) method was developed as a machine learning alternative to parametric logistic regression for detecting interactions in the absence of significant marginal effects. The goal of MDR is to reduce the dimensionality inherent in modelling combinations of polymorphisms using a computational approach called constructive induction. Here, we propose a Robust Multifactor Dimensionality Reduction (RMDR) method that performs constructive induction using a Fisher's Exact Test rather than a predetermined threshold. The advantage of this approach is that only statistically significant genotype combinations are considered in the MDR analysis. We use simulation studies to demonstrate that this approach will increase the success rate of MDR when there are only a few genotype combinations that are significantly associated with case-control status. We show that there is no loss of success rate when this is not the case. We then apply the RMDR method to the detection of gene-gene interactions in genotype data from a population-based study of bladder cancer in New Hampshire.

Epistasis among eNOS, MMP-9 and VEGF maternal genotypes in hypertensive disorders of pregnancy

“…RMDR performs constructive induction using a Fisheŕs exact test rather than a predetermined threshold, and it has the advantage that only statistically significant genotype combinations are considered in the MDR analysis. 24 The cross-validation steps are similar to those of the MDR approach with the exception that a testing score is used to replace the balanced accuracy. The RMDR approach is freely available as part of the open-source MDR software package version 2.0 (http://www.epistasis.org/ 24 ).…”

“…Seven clinically relevant polymorphisms were studied in eNOS (T-786C, rs2070744 in the promoter region; the 27 bp variable number of tandem repeats in intron 4; and Glu298Asp, rs1799983 in exon 7), in the promoter region of MMP-9 (C-1562T, rs3918242 and -90(CA) [13][14][15][16][17][18][19][20][21][22][23][24][25] , rs2234681), and in the promoter region of VEGF (C-2578A, rs699947 and G-634C, rs2010963). Genotypes for eNOS rs2070744 and rs1799983 and for VEGF polymorphisms were determined by TaqMan allele discrimination assays using probes and primers designed by Applied Biosystems (Carlsbad, CA, USA), as previously described.…”

“…Besides the study of these genes independently, epistasis has also been suggested to be an important component of complex diseases, including PE. 17,22,23 Indeed, gene-gene interactions have been successfully used to detect susceptibility to multifactorial diseases, [24][25][26] and it is possible that this analysis could help the genetic basis of PE and GH.…”

“…They have largely focused on maternal genotypes, and polymorphisms in genes involved in angiogenesis, endothelial function and oxidative stress have been associated with PE. [15][16][17] We previously studied clinically relevant polymorphisms of the eNOS (T-786C, 27 bp variable number of tandem repeats in intron 4 and Glu298Asp), MMP-9 (C-1562T and -90(CA) [13][14][15][16][17][18][19][20][21][22][23][24][25] ) and VEGF (C-2578A and G-634C) genes in HDP. We reported associations of some genotypes and haplotypes formed by these polymorphisms with PE and/or gestational hypertension (GH).…”

Clustering subjects in genetic studies with Self Organizing Maps

“…Ritchie et al applied Multifactor Dimensionality Reduction (MDR) to a sporadic breast cancer data set [5], whereas variations of MDR were also introduced for multi-loci associations [6]- [8]. MDR and its variations had good results when they were used on a small number of SNPs but they cannot be directly used on a large number of SNPs, due to the exhaustive search it performs for identifying n-SNP associations [4].…”