Genome-wide association studies have revealed that many single nucleotide polymorphisms (SNPs) are associated with breast cancer, and yet the potential SNP–SNP interactions have not been well addressed to date. This study aims to develop a methodology for the selection of SNP–genotype combinations with a maximum difference between case and control groups. We propose a new chaotic particle swarm optimization (CPSO) algorithm that identifies the best SNP combinations for breast cancer association studies containing seven SNPs. Five scoring functions, that is, the percentage correct, sensitivity/specificity, positive predictive value/negative predictive value, risk ratio, and odds ratio, are provided for evaluating SNP interactions in different SNP combinations. The CPSO algorithm identified the best SNP combinations associated with breast cancer protection. Some SNP interactions in specific SNPs and their corresponding genotypes were revealed. These SNP combinations showed a significant association with breast cancer protection (P<0.05). The sensitivity and specificity of the respective best SNP combinations were all higher than 90%. In contrast to the corresponding non-SNP–SNP interaction combinations, the estimated odds ratio and risk ratio of the SNP–SNP interaction in SNP combinations for breast cancer were less than 100%. This suggests that CPSO can successfully identify the best SNP combinations for breast cancer protection. In conclusion, we focus on developing a methodology for the selection of SNP–genotype combinations with a maximum difference between case and control groups. The CPSO method can effectively identify SNP–SNP interactions in complex biological relationships underlying the progression of breast cancer.
aInstitute of Biotechnology and Chemical Engineering, I-Shou University
bDepartment of Biomedical Science and Environmental Biology, Center of Excellence for Environmental Medicine, Cancer Center, Kaohsiung Medical University Hospital, Kaohsiung Medical University
cDepartment of Electronic Engineering, National Kaohsiung University of Applied Sciences, Kaohsiung
dDepartment of Network Systems, Toko University, Puzih, Chiayi County, Taiwan
Correspondence to Dr Hsueh-Wei Chang, Department of Biomedical Science and Environmental Biology, Kaohsiung Medical University, Kaohsiung 80708, Taiwan Tel: +886 7 312 1101 x 2691; fax: +886 7 312 5339; e-mail: firstname.lastname@example.org Li-Yeh Chuang and Cheng-Hong Yang contributed equally to this work.
Received August 4, 2011
Accepted October 10, 2011