Research Paper Volume 12, Issue 13 pp 13172—13186

An individualized transcriptional signature to predict the epithelial-mesenchymal transition based on relative expression ordering

Tingting Chen1, *, , Zhangxiang Zhao1, *, , Bo Chen1, , Yuquan Wang1, , Fan Yang1, , Chengyu Wang1, , Qi Dong1, , Yaoyao Liu1, , Haihai Liang2, , Wenyuan Zhao1, , Lishuang Qi1, , Yan Xu1, , Yunyan Gu1, ,

  • 1 College of Bioinformatics Science and Technology, Harbin Medical University, Harbin, China
  • 2 College of Pharmacy, Harbin Medical University, Harbin, China
* Equal contribution

Received: February 11, 2020       Accepted: May 25, 2020       Published: July 8, 2020
How to Cite

Copyright © 2020 Chen et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY 3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.


The epithelial-mesenchymal transition (EMT) process is involved in cancer cell metastasis and immune system activation. Hence, identification of gene expression signatures capable of predicting the EMT status of cancer cells is essential for development of therapeutic strategies. However, quantitative identification of EMT markers is limited by batch effects, the platform used, or normalization methods. We hypothesized that a set of EMT-related relative expression orderings are highly stable in epithelial samples yet are reversed in mesenchymal samples. To test this hypothesis, we analyzed transcriptome data for ovarian cancer cohorts from publicly available databases, to develop a qualitative 16-gene pair signature (16-GPS) that effectively distinguishes the mesenchymal from epithelial phenotype. Our method was superior to previous quantitative methods in terms of classification accuracy and applicability to individualized patients without requiring data normalization. Patients with mesenchymal-like ovarian cancer showed poorer overall survival compared to patients with epithelial-like ovarian cancer. Additionally, EMT score was positively correlated with expression of immune checkpoint genes and metastasis. We, therefore, established a robust EMT 16-GPS that is independent of detection platform, batch effects and individual variations, and which represents a qualitative signature for investigating the EMT and providing insights into immunotherapy for ovarian cancer patients.


16-GPS: 16-gene pair signature; AUC: Area under the curve; BH: Benjamini-Hochberg; DEGs: Differentially expressed genes; EMT: Epithelial-mesenchymal transition; FDR: False discovery rate; GEO: Gene Expression Omnibus; ICGC: International Cancer Genome Consortium; OvCa: Ovarian cancer; REOs: Relative expression orderings; ROC: Receiver operating characteristic; TCGA: The Cancer Genome Atlas; TGF-β1: Transforming growth factor-β1.