Publication: Searching single nucleotide polymorphism markers to complex diseases using genetic algorithm framework and a BoostMode support vector machine
Issued Date
2010-09-06
Resource Type
Other identifier(s)
2-s2.0-77956138904
Rights
Mahidol University
Rights Holder(s)
SCOPUS
Bibliographic Citation
2010 4th International Conference on Bioinformatics and Biomedical Engineering, iCBBE 2010. (2010)
Suggested Citation
Khantharat Anekboon, Suphakant Phimoltares, Chidchanok Lursinsap, Sissades Tongsima, Suthat Fucharoen Searching single nucleotide polymorphism markers to complex diseases using genetic algorithm framework and a BoostMode support vector machine. 2010 4th International Conference on Bioinformatics and Biomedical Engineering, iCBBE 2010. (2010). doi:10.1109/ICBBE.2010.5515780 Retrieved from: https://repository.li.mahidol.ac.th/handle/20.500.14594/29074
Research Projects
Organizational Units
Authors
Journal Issue
Thesis
Title
Searching single nucleotide polymorphism markers to complex diseases using genetic algorithm framework and a BoostMode support vector machine
Abstract
With the advent of large-scale high density single nucleotide polymorphism (SNP) arrays, case-control association studies have been performed to identify predisposing genetic factors that influence many common complex diseases. These genotyping platforms provide very dense SNP coverage per one chip. Much research has been focusing on multivariate genetic model to identify genes that can predict the disease status. However, increasing the number of SNPs generates large number of combined genetic outcomes to be tested. This work presents a new mathematical algorithm for SNP analysis called IFGA that uses a "BoostMode" support vector machine (SVM) to select the best set of SNP markers that can predict a state of complex diseases. The proposed algorithm has been applied to test for the association study in two diseases, namely Crohn's and severity spectrum of β0°/Hb E Thalassemia diseases. The results revealed that our predicted SNPs can respectively best classify both diseases at 71.57% and 71.06% accuracy using 10-fold cross validation comparing with the optimum random forest (ORF) and classification and regression trees (CART) techniques. © 2010 IEEE.