Comparison of evolutionary algorithms in gene regulatory network model inference.
Affiliation
Centre for Scientific Computing and Complex Systems Modelling, Dublin City University, Dublin 9, Ireland. asirbu@computing.dcu.ie.Issue Date
2010
Metadata
Show full item recordCitation
Comparison of evolutionary algorithms in gene regulatory network model inference. 2010, 11:59 BMC BioinformaticsJournal
BMC bioinformaticsDOI
10.1186/1471-2105-11-59PubMed ID
20105328Abstract
ABSTRACT: BACKGROUND: The evolution of high throughput technologies that measure gene expression levels has created a data base for inferring GRNs (a process also known as reverse engineering of GRNs). However, the nature of these data has made this process very difficult. At the moment, several methods of discovering qualitative causal relationships between genes with high accuracy from microarray data exist, but large scale quantitative analysis on real biological datasets cannot be performed, to date, as existing approaches are not suitable for real microarray data which are noisy and insufficient. RESULTS: This paper performs an analysis of several existing evolutionary algorithms for quantitative gene regulatory network modelling. The aim is to present the techniques used and offer a comprehensive comparison of approaches, under a common framework. Algorithms are applied to both synthetic and real gene expression data from DNA microarrays, and ability to reproduce biological behaviour, scalability and robustness to noise are assessed and compared. CONCLUSIONS: Presented is a comparison framework for assessment of evolutionary algorithms, used to infer gene regulatory networks. Promising methods are identified and a platform for development of appropriate model formalisms is established.Language
enISSN
1471-2105ae974a485f413a2113503eed53cd6c53
10.1186/1471-2105-11-59