BMC Bioinformatics
Research article
BioMed Central
Open Access
Disease candidate gene identification and prioritization using
protein interaction networks
JingChen1,2, BruceJAronow1,2,3 and AnilGJegga*1,3
Address: 1Division of Biomedical Informatics, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, USA, 2Department of Biomedical Engineering, University of Cincinnati, Cincinnati, OH, USA and 3Department of Pediatrics, University of Cincinnati College of Medicine, Cincinnati, OH, USA
Email: JingChen-Jing.Chen@cchmc.org; BruceJAronow-Bruce.Aronow@cchmc.org; AnilGJegga*-Anil.Jegga@cchmc.org* Corresponding author
Published: 27 February 2009BMC Bioinformatics 2009, 10:73
doi:10.1186/1471-2105-10-73
This article is available from: doc.guandang.net
Received: 2 June 2008
Accepted: 27 February 2009
© 2009 Chen et al; licensee BioMed Central Ltd.
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Abstract
Background: Although most of the current disease candidate gene identification andprioritization methods depend on functional annotations, the coverage of the gene functionalannotations is a limiting factor. In the current study, we describe a candidate gene prioritizationmethod that is entirely based on protein-protein interaction network (PPIN) analyses.
Results: For the first time, extended versions of the PageRank and HITS algorithms, and the K-Step Markov method are applied to prioritize disease candidate genes in a training-test schema.Using a list of known disease-related genes from our earlier study as a training set ("seeds"), andthe rest of the known genes as a test list, we perform large-scale cross validation to rank thecandidate genes and also evaluate and compare the performance of our approach. Underappropriate settings – for example, a back probability of 0.3 for PageRank with Priors and HITSwith Priors, and step size 6 for K-Step Markov method – the three methods achieved a comparableAUC value, suggesting a similar performance.
Conclusion: Even though network-based methods are generally not as effective as integratedfunctional annotation-based methods for disease candidate gene prioritization, in a one-to-onecomparison, PPIN-based candidate gene prioritization performs better than all other gene featuresor annotations. Additionally, we demonstrate that methods used for studying both social and Webnetworks can be successfully used for disease candidate gene prioritization.
Background
Most of the current disease candidate gene prioritizationmethods [1-6] rely on functional annotations. However,the coverage of the gene functional annotations is a limit-ing factor. Although more than 1,500 human diseasegenes have been documented, most of them continue tobe functionally uncharacterized. Currently, only a fractionof the genome is annotated with pathways and pheno-types [6]. While two thirds of all the genes are annotated
by at least one functional annotation, the remaining onethird is yet to be annotated.
Analysis of protein-protein interaction networks (PPINs)is important for inferring the function of uncharacterizedproteins. Protein-protein interactions refer to the associa-tion among the protein molecules and the study of theseassociations from the perspective of biochemistry, signaltransduction and biomolecular networks. Recent biotech-
Prioritization Jianghui Xiong 熊江辉 Laserxiong@gmail... 2005 What’s wrong with our Disease Models The...Candidate Module C Query Module A Whole Gene Set...
热门文档