A0345
Title: A novel statistical framework for case-control genome-wide association studies
Authors: Itziar Irigoien - University Basque Country (Spain) [presenting]
Selena Aranda - Universitat de Barcelona (Spain)
Marina Mitjans - Universitat de Barcelona (Spain)
Bru Cormand - Universitat de Barcelona (Spain)
Concepcion Arenas - University of Barcelona (Spain)
Group Cibersam - Spanish Mental Health Research Network (Spain)
Abstract: Genome-wide association studies (GWAS) investigate the relationship between genetic variation and traits of interest by analyzing molecular markers distributed throughout the genome, such as single-nucleotide polymorphisms (SNPs). The use of GWAS has been growing over the years, enabling the identification of genetic loci associated with a wide range of phenotypes, including asthma, diabetes, and psychiatric disorders. In case-control designs, the most commonly used statistical method is logistic regression, which typically includes as covariates the principal components derived from a multidimensional scaling (MDS) analysis to account for population stratification. The optimal number of components to include depends on the population structure and sample size, although the inclusion of up to 10 components is generally accepted. Given that many SNPs are correlated due to linkage disequilibrium, the Bonferroni correction for multiple testing is often overly conservative, leading to an increased risk of false negative findings. To address this limitation, a multivariate statistical approach is proposed based on distances and k-neighborhoods, which offers advantages in detecting associations in case-control GWAS. The application of this method is illustrated using real data, highlighting both its implementation and the results it provides.
