GWAS works backwards in relationship on how pathophysiology is discovered. Given a big dataset of genes, we could do a cross-section of mutations and see the SNPs present in patients with a disease, study the mutations and determine if it’s protective or a risk factor for a disease. Then the loci is studied and make inferences on how the alteration on how the molecular physiology is altered. When the variant is non-coding, it’s more likely that gene expression is altered by disrupting transcription binding motifs or altering chromating interactions.