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

Table 3

Summary of detect gene-gene interaction using random forest method.

No.	Author	Dataset	Description
(1)	Lunetta et al. [33]	H2M2, H4M2, H8M2, H16M2, H4M4, and H8M4.	RF as a screening procedure to identify top-ranked true-associated SNPs which can cause disease without losing any interactions.
(2)	Jiang et al. [34]	Three simulated disease model.	RF is used to recognize the cases that were against controls and to obtain the Gini importance which is used to measure the contribution of each SNP to the classification performance.
(3)	Schwarz et al. [35]	Crohn’s disease.	A new method of RJ based on basis RF knowledge was developed to facilitate a fast processing in the high-dimensional of genome-wide analysis data of gene-gene interactions.
(4)	Liu et al. [36]	NARAC1 and NARAC2.	RF is used to detect contributed gene-gene interactions for identifing RA susceptibility and to identify SNPs of RA patients to classify them into anticyclic citrullinated protein positive and healthy controls.
(5)	Winham et al. [32]	Five models.	Focus on identifing rarely gene-gene interactions and detecting gene-gene interaction effects and their potential effectiveness on high-dimensional data using RF.
(6)	Pan et al. [37]	Bladder cancer.	The proposed method of MINGRF is proposed to improve the performance of RF such as accuracy and computational time.
(7)	Staiano et al. [38]	Familial combined hyperlipidemia (FCH).	RF is used to identify gene-gene interactions that are involved in FCH. FCH increase the plasma triglycerides and/or total cholesterol level of patients and hence increase the risk of coronary heart disease.
(8)	Chen and Ishwaran [39]	Colon cancer and ovarian cancer.	RSF as new hunting pathway to detect gene correlation and genomic interactions from a high-dimensional genomic data.