Identifying genes with tri-modal association with survival and tumor grade in cancer patients

Minzhe Zhang,Tao Wang,Rosa Sirianni,Yang Xie,Philip W Shaul

doi:10.1186/s12859-018-2582-7

Abstract

BackgroundPrevious cancer genomics studies focused on searching for novel oncogenes and tumor suppressor genes whose abundance is positively or negatively correlated with end-point observation, such as survival or tumor grade. This approach may potentially miss some truly functional genes if both its low and high modes have associations with end-point observation. Such genes act as both oncogenes and tumor suppressor genes, a scenario that is unlikely but theoretically possible.ResultsWe invented an Expectation-Maximization (EM) algorithm to divide patients into low-, middle- and high-expressing groups according to the expression level of a certain gene in both tumor and normal patients. We found one gene, ORMDL3, whose low and high modes were both associated with worse survival and higher tumor grade in breast cancer patients in multiple patient cohorts. We speculate that its tumor suppressor gene role may be real, while its high expression correlating with worse end-point outcome is probably due to the passenger event of the nearby ERBB2’s amplification.ConclusionsThe proposed EM algorithm can effectively detect genes having tri-modal distributed expression in patient groups compared to normal genes, thus rendering a new perspective on dissecting the association between genomic features and end-point observations. Our analysis of breast cancer datasets suggest that the gene ORMDL3 may have an unexploited tumor suppressive function.

Highlights

Previous cancer genomics studies focused on searching for novel oncogenes and tumor suppressor genes whose abundance is positively or negatively correlated with end-point observation, such as survival or tumor grade
We focused on a specific scenario where candidate targets whose “low” and “high” groups of patients were both associated with worse survival and higher tumor grade compared to the “middle” group of patients
Grouping of patients into 3 modes by EM algorithm We focused on the cases where the tumor patients can be grouped into “low”, “middle” and “high” groups according to expression of a certain gene

Summary

Introduction

Previous cancer genomics studies focused on searching for novel oncogenes and tumor suppressor genes whose abundance is positively or negatively correlated with end-point observation, such as survival or tumor grade. Research in cancer biology has identified hundreds of genes involved in different stages of tumorigenesis [7, 17] The alterations in these oncogenes or tumor suppressor genes can come from a variety of sources, such as single nucleotide polymorphisms (SNPs), copy number variations (CNV), chromosomal regions, viral integration, gene fusions, etc. There is another type of event called a passenger mutation, which commonly occurs in tumor tissues. Bric et al conducted an RNA interference (RNAi) screen for

Methods

Results

Discussion

Conclusion