A novel unsupervised algorithm for biological process-based analysis on cancer

Tianci Song, Sha Cao, Sheng Tao, Sen Liang, Wei Du, Yanchun Liang

Research output: Contribution to journalArticle

1 Scopus citations

Abstract

The aberrant alterations of biological functions are well known in tumorigenesis and cancer development. Hence, with advances in high-throughput sequencing technologies, capturing and quantifying the functional alterations in cancers based on expression profiles to explore cancer malignant process is highlighted as one of the important topics among cancer researches. In this article, we propose an algorithm for quantifying biological processes by using gene expression profiles over a sample population, which involves the idea of constructing principal curves to condense information of each biological process by a novel scoring scheme on an individualized manner. After applying our method on several large-scale breast cancer datasets in survival analysis, a subset of these biological processes extracted from corresponding survival model is then found to have significant associations with clinical outcomes. Further analyses of these biological processes enable the study of the interplays between biological processes and cancer phenotypes of interest, provide us valuable insights into cancer biology in biological process level and guide the precision treatment for cancer patients. And notably, prognosis predictions based on our method are consistently superior to the existing state of art methods with the same intention.

Original languageEnglish (US)
Article number4671
JournalScientific reports
Volume7
Issue number1
DOIs
StatePublished - Dec 1 2017

    Fingerprint

ASJC Scopus subject areas

  • General

Cite this