Identification of molecular biomarkers for pancreatic cancer with mRMR shortest path method
Metrics: PDF 1576 views | HTML 2007 views | ?
Shuhua Shen1, Tuantuan Gui2 and Chengcheng Ma3,4
1Zhejiang Provincial Hospital of Traditional Chinese Medicine, Hangzhou, China
2Shanghai Smartquerier Biotechnology Co., Ltd, Shanghai, China
3CAS-MPG Partner Institute for Computational Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China
4Shanghai Center for Bioinformatics Technology, Shanghai, China
Chengcheng Ma, email: firstname.lastname@example.org
Keywords: minimum-redundancy-maximum-relevance (mRMR), pancreatic cancer, biomarker
Received: February 10, 2017 Accepted: April 20, 2017 Published: May 25, 2017
The high mortality rate of pancreatic cancer makes it one of the most studied diseases among all cancer types. Many researches have been conducted to understand the mechanism underlying its emergence and pathogenesis of this disease. Here, by using minimum-redundancy-maximum-relevance (mRMR) method, we studied a set of transcriptome data of pancreatic cancer. As we gradually added features to achieve the most accurate classification results of Jackknife, a gene set of 9 genes was identified. They were NHS, SCML2, LAMC2, S100P, COL17A1, AMIGO2, PTPRR, KPNA7 and KCNN4. Through STRING 2.0 protein-protein interactions (PPIs) analysis, 40 proteins were identified in the shortest paths between genes in the gene set, 30 of them passed the permutation test, which indicated they were hubs in the background network. Those genes in the protein-protein interaction network were enriched to 37 functional modules, such as: negative regulation of transcription from RNA polymerase II promoter, negative regulation of ERK1 and ERK2 cascade and BMP signaling pathway. Our study indicated new mechanism of pancreatic cancer, suggesting potential therapeutic targets for further study.
All site content, except where otherwise noted, is licensed under a Creative Commons Attribution 3.0 License.