Prediction of a time-to-event trait using genome wide SNP data

Predictive and prognostic molecular markers for cholangiocarcinoma in Han Chinese population

Cholangiocarcinoma (CCA) is the most common malignant heterogeneous polygenetic carcinoma with a high incidence in Asia. Most patients would die within 1 year after diagnosis and the 5 year survival rate is less than 10-20% worldwide. Single nucleotide polymorphisms (SNPs) in genes regulate telomere maintenance, mitosis, and inflammation, and may help predict individual susceptibility to certain drugs, environmental factor, and risks to particular diseases. The gene-gene interaction and the regulation of SNPs have not been assessed extensively in CCA. According to our previous study, the GRB2-associated-binding protein (Gab1) gene rs3805246 (X(2) =5.015, P=0.025, OR=0.531, 95% CI 0.304-0.928) and epidermal growth factor receptor (EGFR) gene rs2007000 (X(2) =7.934, P=0.005, OR=2.148, 95% CI 1.255-3.675) presented significant difference between CCA patients and controls. This study conducted a population-based analysis using 225 CCA cases (153 biliary tract cancer patients and 72 gall bladder cancer patients) to assess the association between SNPs and progression of CCA patients, including the overall survival and the prognosis analysis. Results showed that an increased susceptibility of BTC was significantly associated with SNP loci distribution frequency in EGFR rs2107000 (X(2) =7.934, P=0.005, OR=2.148, 95% CI 1.255-3.675). Furthermore, multivariate factor regression analysis represented cholelithiasis medical history of BTC patients can be an effective evaluation criteria of BTC susceptibility in early stage. This study also assessed the relationship between these genotypic polymorphisms and clinicopathologic data, including tumor differentiation stage and overall survival. This is the first study identifying that EGFR polymorphisms are associated with BTC and EGFR rs2017000 polymorphisms may be an important survival predictor in BTC patients.

SNP selection in genome-wide association studies via penalized support vector machine with MAX test

One of main objectives of a genome-wide association study (GWAS) is to develop a prediction model for a binary clinical outcome using single-nucleotide polymorphisms (SNPs) which can be used for diagnostic and prognostic purposes and for better understanding of the relationship between the disease and SNPs. Penalized support vector machine (SVM) methods have been widely used toward this end. However, since investigators often ignore the genetic models of SNPs, a final model results in a loss of efficiency in prediction of the clinical outcome. In order to overcome this problem, we propose a two-stage method such that the the genetic models of each SNP are identified using the MAX test and then a prediction model is fitted using a penalized SVM method. We apply the proposed method to various penalized SVMs and compare the performance of SVMs using various penalty functions. The results from simulations and real GWAS data analysis show that the proposed method performs better than the prediction methods ignoring the genetic models in terms of prediction power and selectivity.