Single nucleotide polymorphism selection using independent component analysis

Bioinformatics research in genome wide association studies necessitates the development of algorithms capable of manipulating very-large datasets of Single Nucleotide Polymorphisms (SNP). To facilitate such association studies, we propose a novel framework for SNP selection using Independent Component Analysis (ICA). Compared to previous ICA-based methods, our framework works as a filtering technique to reduce the number of SNPs in a dataset, without the need for any class labels. We evaluate the proposed method by applying it on three published SNP datasets, and comparing the results to SNP selection methods based on Principal Component Analysis (PCA). Our results show the capability of ICA to capture an increased or matching amount of information from the datasets.