Mining time-dependent gene features

Research output: Contribution to journalArticlepeer-review

6 Scopus citations

Abstract

This paper presents an application of the Independent Component Analysis (ICA) method to genomic data. In particular, experimentally produced perturbation effects over the E.coli bacterium are monitored through the changes of gene expression values observed at regular times, and until steady state has been reached. The aim is to control the response of the SOS system to DNA damage. We might assume that only part of the genetic regulatory network is affected directly by the perturbation conditions, as indirect cascade effects might also be present, and some genes may change just because of randomness. ICA decomposes the gene matrix and identifies groups of genes belonging to a certain estimated component by virtue of co-expression; it is of course of interest to establish co-regulation dynamics, which might underlie the captured correlation. Stronger forms of dependence, like Mutual Information, are thus computed and compared with linear correlation in order to validate the results and establish the role of the identified components in determining the network dynamics.

Original languageEnglish (US)
Pages (from-to)1191-1205
Number of pages15
JournalJournal of Bioinformatics and Computational Biology
Volume3
Issue number5
DOIs
StatePublished - Oct 2005
Externally publishedYes

Keywords

  • Correlation and dependence
  • Genomics
  • ICA
  • Mutual information

ASJC Scopus subject areas

  • Biochemistry
  • Molecular Biology
  • Computer Science Applications

Fingerprint

Dive into the research topics of 'Mining time-dependent gene features'. Together they form a unique fingerprint.

Cite this