Method of joint clustering in network and correlation spaces

Anastasiia N. Gainullina, Maxim Artyomov, Alexey A. Sergushichev

Research output: Contribution to journalArticlepeer-review

Abstract

Subject of Research. The joint clustering method in network and correlation context is designed to identify active modules in metabolic graphs based on transcriptomic data represented by a large number of samples. The active modules obtained by this method describe the dynamic metabolic regulation in all samples of the analyzed dataset. The paper presents modifications of the proposed method for application on real data. Method. For results stability study the modified method was repeatedly run on real data with small variations of the initial parameters. For result analysis, several metrics were formulated that display modules similarity and representation under various start-up conditions. Main Results. The analysis results are sufficiently robust. For the most modules, their profiles are detected well in the noisy data, and the most genes are also preserved. Practical Relevance. The results of the presented study have shown that the modified method analyzes successfully real data by producing active modules that are stable and easy in interpretation.

Original languageEnglish
Pages (from-to)807-814
Number of pages8
JournalScientific and Technical Journal of Information Technologies, Mechanics and Optics
Volume20
Issue number6
DOIs
StatePublished - Nov 1 2020

Keywords

  • Clustering
  • Correlation
  • Gene expression
  • Graphs
  • Metabolic networks
  • Transcriptomic data

Fingerprint

Dive into the research topics of 'Method of joint clustering in network and correlation spaces'. Together they form a unique fingerprint.

Cite this