Network approaches to Genome-Wide Association studies

In the framework of large-scale genotypic studies (describing the distribution of allele frequencies inside human genome) we characterize the Linkage Disequilibrium (LD) matrix as a network of relationships between alleles. We propose a suitable matrix discretization threshold, after a characterization of the distribution of noisy values inside LD matrix. We compare the main network parameters of a real LD matrix with two null models (Erdos-Renyi random network and a rewiring of the original network), in order to highlight the peculiar features of the LD network. We conclude stating the need of adequate computing tools for handling the high-dimensional data coming from Genome-Wide genotyping datasets.