Detection of SNP-SNP interaction based on the generalized particle swarm optimization algorithm

Most of complex diseases are believed to be mainly caused by epistatic interactions of pair single nucleotide poly-morphisms (SNPs), namely, SNP-SNP interactions. Though many works have been done for the detection of SNP-SNP interactions, the algorithmic development is still ongoing due to their mathematical and computational complexities. In this study, we proposed a method, PSOMiner, based on the generalized particle swarm optimization algorithm, with mutual information as its fitness function, for the detection of SNP-SNP interaction that has the highest pathogenic effect in a SNP data set. Experiments of PSOMiner are performed on six simulation data sets under the criteria of detection power. Results demonstrate that PSOMiner is promising for the detection of SNP-SNP interaction. In addition, the application of PSOMiner on a real age-related macular degeneration (AMD) data set provides several new clues for the exploration of AMD associated SNPs that have not been described previously. PSOMiner might be an alternative to existing methods for detecting SNP-SNP interactions.