GWAS analysis to compute genetic markers of progression to Alzheimer's disease

Prediction of conversion from Mild Cognitive Impairment (MCI) to Alzheimer's Disease (AD) is a challenging task due to involvement of many genetic and environmental factors leading to neurodegeneration. Several Genome Wide Association Studies (GWAS) have been conducted to understand genetic risk factors of AD, however, there are not many GWAS studies that find the genetic factors of conversion from MCI to AD. In this study, we aim to find potential genetic markers causing conversion from MCI to AD and build a machine learning model to predict the conversion. To this end, we used genetic variation data of 809 patients in Alzheimer's Disease Neuroimaging Initiative (ADNI). We processed the genetic data by merging SNPs information of all patients into a single set of PLINK files, which served as input for the GWAS analysis. SNPs were filtered based on call rate, minor allele frequency and Hardy-Weinberg equilibrium cut-off. Samples were also filtered to remove any linkage using sample call rate, inbreeding coefficient, kinship coefficient and linkage disequilibrium cut-off. After the filtration, we calculated negative log p-value for the remaining SNPs in association with the phenotype (i.e., converter vs. non-converter). We repeated this step for other traits such as ventricle size, hippocampus size, and Tau protein concentration to record different loci on chromosomes influencing each trait. The presence or absence of significant SNPs will be used to create a deep learning model to predict the progression of AD in a patient using only genetic information.