Learning Gene Regulatory Networks Computationally from Gene Expression Data Using Weighted Consensus

Handle URI:
http://hdl.handle.net/10754/550417
Title:
Learning Gene Regulatory Networks Computationally from Gene Expression Data Using Weighted Consensus
Authors:
Fujii, Chisato ( 0000-0002-9762-7592 )
Abstract:
Gene regulatory networks analyze the relationships between genes allowing us to un- derstand the gene regulatory interactions in systems biology. Gene expression data from the microarray experiments is used to obtain the gene regulatory networks. How- ever, the microarray data is discrete, noisy and non-linear which makes learning the networks a challenging problem and existing gene network inference methods do not give consistent results. Current state-of-the-art study uses the average-ranking-based consensus method to combine and average the ranked predictions from individual methods. However each individual method has an equal contribution to the consen- sus prediction. We have developed a linear programming-based consensus approach which uses learned weights from linear programming among individual methods such that the methods have di↵erent weights depending on their performance. Our result reveals that assigning di↵erent weights to individual methods rather than giving them equal weights improves the performance of the consensus. The linear programming- based consensus method is evaluated and it had the best performance on in silico and Saccharomyces cerevisiae networks, and the second best on the Escherichia coli network outperformed by Inferelator Pipeline method which gives inconsistent results across a wide range of microarray data sets.
Advisors:
Gao, Xin ( 0000-0002-7108-3574 )
Committee Member:
Soloviev, Victor; Hoehndorf, Robert ( 0000-0001-8149-5890 )
KAUST Department:
Computer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division
Program:
Computer Science
Issue Date:
16-Apr-2015
Type:
Thesis
Appears in Collections:
Theses; Computer Science Program; Computer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division

Full metadata record

DC FieldValue Language
dc.contributor.advisorGao, Xinen
dc.contributor.authorFujii, Chisatoen
dc.date.accessioned2015-04-21T11:37:58Zen
dc.date.available2015-04-21T11:37:58Zen
dc.date.issued2015-04-16en
dc.identifier.urihttp://hdl.handle.net/10754/550417en
dc.description.abstractGene regulatory networks analyze the relationships between genes allowing us to un- derstand the gene regulatory interactions in systems biology. Gene expression data from the microarray experiments is used to obtain the gene regulatory networks. How- ever, the microarray data is discrete, noisy and non-linear which makes learning the networks a challenging problem and existing gene network inference methods do not give consistent results. Current state-of-the-art study uses the average-ranking-based consensus method to combine and average the ranked predictions from individual methods. However each individual method has an equal contribution to the consen- sus prediction. We have developed a linear programming-based consensus approach which uses learned weights from linear programming among individual methods such that the methods have di↵erent weights depending on their performance. Our result reveals that assigning di↵erent weights to individual methods rather than giving them equal weights improves the performance of the consensus. The linear programming- based consensus method is evaluated and it had the best performance on in silico and Saccharomyces cerevisiae networks, and the second best on the Escherichia coli network outperformed by Inferelator Pipeline method which gives inconsistent results across a wide range of microarray data sets.en
dc.language.isoenen
dc.subjectgene regulatory networksen
dc.subjectconsensusen
dc.subjectlinear programmingen
dc.titleLearning Gene Regulatory Networks Computationally from Gene Expression Data Using Weighted Consensusen
dc.typeThesisen
dc.contributor.departmentComputer, Electrical and Mathematical Sciences and Engineering (CEMSE) Divisionen
thesis.degree.grantorKing Abdullah University of Science and Technologyen_GB
dc.contributor.committeememberSoloviev, Victoren
dc.contributor.committeememberHoehndorf, Roberten
thesis.degree.disciplineComputer Scienceen
thesis.degree.nameMaster of Scienceen
dc.person.id129135en
All Items in KAUST are protected by copyright, with all rights reserved, unless otherwise indicated.