A Review for Detecting Gene-Gene Interactions Using Machine Learning Methods in Genetic Epidemiology

Table 2

Summary of detect gene-gene interaction using support vector machine method.

No.	Author	Dataset	Description
(1)	Matchenko-Shimko and Dubé [23]	Simulated disease.	Both SVM and artificial neural network (ANN) were used to preselect the combination of SNP to test the importance of potential interactions between genes in complex disease.
(2)	Chen et al. [19]	Real prostate cancer genotyping.	SVM was applied in different kinds of combinatorial optimization methods which were recursive feature addition, recursive feature elimination, local search, and genetic algorithm.
(3)	Özgür et al. [24]	Prostate cancer.	Automatic method that was proposed to extract known genes-disease and infer unknown gene-disease association by using automatic literature mining based on dependency parsing and support vector machines.
(4)	Shen et al. [25]	Parkinson disease.	Authors had employ two-stage method by using SVM with L1 penalty to detect gene-gene interactions for human complex disease.
(5)	Ban et al. [26]	Type 2 diabetes mellitus-related genes.	SVM was used to predict the importance of gene-gene interactions in T2D in the studies of Korean cohort studies.
(6)	Missiuro [21]	Caenorhabditis elegans.	SVM was utilized in this research to detect interactions between gene in kinase families for Caenorhabditis elegans organism.
(7)	Fang and Chiu [27]	COGA (genetics of alcoholism).	SVM-based PGMDR was introduced to study the interactions of gene-gene and gene-covariate in the presence or absence of main effects of genes.
(8)	Zhang et al. [28]	Human cancer.	Binary matrix shuffling filter (BMSF) as an efficient SVM search schemes was integrated with SVM to classify cancer tissue samples.
(9)	Marvel and Motsinger-Reif [29]	Disease model, M1 and M2.	GESVM was applied in large dataset to select important features, parameters, or kernel in SVM.