Abstract:
|
Analysis of genome-wide data is riddled with unprecedented challenges because of the newest second generation sequencing techniques. A novel statistical approach, DC-RR, has recently been developed to inspect the genome-wide associations after overcoming high dimensionality and the complex correlation structure present in the whole genome data. The DC-RR approach selects the most promising SNPs by removing noise using a Distance Correlation based feature screening approach. The correlation structure is then aggressively addressed using the ridge penalized multiple logistic regression model. The DC-RR method has demonstrated a gain in power and reduced false discovery rate relative to other popular methods. In this article, we implemented DC-RR approach to detect susceptibility gene for Polycystic ovary (PCOS) disease. After analyzing the dataset that has 1042 polycystic ovary patients and 3055 controls over more than 700,000 SNPs, a few significant SNPs were located. These findings will better assist researchers in public health for diagnosis and treatment of the PCOS disease.
|