Identification and characterization of genes with absolute mRNA abundances changes in tumor cells with varied transcriptome sizes

Hao Cai,You Guo,Zheng Guo,Qingzhou Guan,Haidan Yan,Jun He,Wenbin Zhou,Xiangyu Li,Xianlong Wang,Huaping Liu,Kai Song

doi:10.1186/s12864-019-5502-y

Abstract

BackgroundThe amount of RNA per cell, namely the transcriptome size, may vary under many biological conditions including tumor. If the transcriptome size of two cells is different, direct comparison of the expression measurements on the same amount of total RNA for two samples can only identify genes with changes in the relative mRNA abundances, i.e., cellular mRNA concentration, rather than genes with changes in the absolute mRNA abundances.ResultsOur recently proposed RankCompV2 algorithm identify differentially expressed genes (DEGs) through comparing the relative expression orderings (REOs) of disease samples with that of normal samples. We reasoned that both the mRNA concentration and the absolute abundances of these DEGs must have changes in disease samples. In simulation experiments, this method showed excellent performance for identifying DEGs between normal and disease samples with different transcriptome sizes. Through analyzing data for ten cancer types, we found that a significantly higher proportion of the DEGs with absolute mRNA abundance changes overlapped or directly interacted with known cancer driver genes and anti-cancer drug targets than that of the DEGs only with mRNA concentration changes alone identified by the traditional methods. The DEGs with increased absolute mRNA abundances were enriched in DNA damage-related pathways, while DEGs with decreased absolute mRNA abundances were enriched in immune and metabolism associated pathways.ConclusionsBoth the mRNA concentration and the absolute abundances of the DEGs identified through REOs comparison change in disease samples in comparison with normal samples. In cancers these genes might play more important upstream roles in carcinogenesis.

Highlights

The amount of RNA per cell, namely the transcriptome size, may vary under many biological conditions including tumor
We provided preliminary evidence that the differentially expressed genes (DEGs) with changes in both absolute mRNA abundances and concentration are more likely to be closely related with cancer driver genes and drug targets than the DEGs which may change only in mRNA concentration exclusively identified by the popular SAM or edgeR algorithm
With the 616 cancer driver genes downloaded from the Catalogue Of Somatic Mutations (COSMIC, version81, updated 9th May 2017) database [33], we found 25.79% of the 6472 absolute DEGs of liver hepatocellular carcinoma (LIHC) overlapped or directly interacted with known cancer driver genes based on the protein–protein interaction data downloaded from the STRING v10 database [34], which was significantly higher than the corresponding ratio (18.35%) for the 12,049 relative DEGs (Fisher’s exact test, p < 1.0E-16)

Summary

Introduction

The amount of RNA per cell, namely the transcriptome size, may vary under many biological conditions including tumor. It is a common practice to identify differentially expressed genes (DEGs) between two phenotypes through comparing the gene expression profiles measured with the same amount of RNA (or mRNA) extracted from two-phenotype samples, based on the assumption that different types of cells have approximately the same amount of total RNA per cell (transcriptome size) [1]. Cai et al BMC Genomics (2019) 20:134 it could be argued that the concentrations of macromolecules are relevant parameters governing biochemical reactions inside cells, inappropriate interpretation of mRNA concentration changes might lead to incorrect conclusions for a range of biological questions, including the transcriptional characteristics of cancer cells [7]

Methods

Results

Discussion

Conclusion