Identification of genes related to cancer through targeted sequencing data analysis

Abstract

Recent statistical methods for next generation sequencing (NGS) data have been successfully applied to identifying rare genetic variants associated with certain diseases. Note that most commonly used methods such as burden tests and variance-component tests rely on large sample size. However, due to a high cost of sequencing, small sample size sequencing data are popularly generated. Most existing methods are not appropriate to handle sequencing data with small samples. In this work, we propose a new exact association test for sequencing data which does not require a large sample approximation. Our method is based upon the generalized Cochran-Mantel-Haenszel (CMH) statistic. We applied our method to NGS data from Intraductal papillary mucinous neoplasm (IPMN) patients. These IPMN patients have the unique pancreatic neoplasm which could turn into an invasive and hard-to-treat pancreatic cancer. Through this application, we successfully identified susceptible genes that are associated with the progression of IPMN to pancreatic cancer.