Optimization of Smoothing Parameters in Splines in GWAS Using a Replication Strategy

Optimizing parameters for spline regression and analyzing extensive data within limited time and resources pose significant challenges, particularly in the context of Genome-Wide Association Studies (GWAS). This research aims to identify associations between genomic variants and phenotypes, focusing on the relationships between groups of single nucleotide polymorphisms (SNPs) and disease risk, as identified by the Sequence Kernel Association Test (SKAT). Spline regression is employed to smooth the p-values while minimizing the error rate. The objectives of this research include optimizing the smoothing parameters for spline regression and determining the optimal number of replications for simulated data, with a particular focus on identifying gene regions potentially associated with Crohn's disease. The findings indicate that a degree of freedom of 1,000 optimizes the model's efficiency, while 1,000 replications yield reasonable outcomes with minimal processing time. Ultimately, the spline regression model with a degree of freedom of 1,000 successfully identifies gene regions associated with Crohn's disease, notably the NOD2 gene, which is widely recognized as relevant to the disease.