The ability to measure the transcriptional response after a stimulus has drawn much attention to the underlying gene regulatory networks. Several machine learning related methods, such as Bayesian networks and decision trees, have been proposed to deal with this difficult problem, but rarely a systematic comparison between different algorithms has been performed. In this work, we critically evaluate the application of multiple linear regression, SVMs, decision trees and Bayesian networks to reconstruct the budding yeast cell cycle network. The performance of these methods is assessed by comparing the topology of the reconstructed models to a validation network. This validation network is defined a priori and each interaction is specified by at least one publication. We also investigate the quality of the network reconstruction if a varying amount of gene regulatory dependencies is provided a priori.
@inproceedings{Supper2007, author = {Supper, Jochen and Fr\"ohlich, Holger and Spieth, Christian and Dr\"ager, Andreas and Zell, Andreas}, title = {Inferring Gene Regulatory Networks by Machine Learning Methods}, booktitle = {Proceedings of the 5\textsuperscript{th} Asia-Pacific Bioinformatics Conference (APBC 2007)}, year = {2007}, editor = {Sankoff, David and Wang, Lusheng and Chin, Francis}, volume = {5}, series = {Series on Advances in Bioinformatics and Computational Biology}, pages = {247--256}, address = {57 Shelton Street, Govent Garden, London WC2H 9HE, UK}, month = jan, publisher = {Imperial College Press}, abstract = {The ability to measure the transcriptional response after a stimulus has drawn much attention to the underlying gene regulatory networks. Several machine learning related methods, such as Bayesian networks and decision trees, have been proposed to deal with this difficult problem, but rarely a systematic comparison between different algorithms has been performed. In this work, we critically evaluate the application of multiple linear regression, SVMs, decision trees and Bayesian networks to reconstruct the budding yeast cell cycle network. The performance of these methods is assessed by comparing the topology of the reconstructed models to a validation network. This validation network is defined \emph{a priori} and each interaction is specified by at least one publication. We also investigate the quality of the network reconstruction if a varying amount of gene regulatory dependencies is provided \emph{a priori}.}, doi = {10.1142/9781860947995_0027}, pdf = {http://www.cogsys.cs.uni-tuebingen.de/publikationen/2007/SupperAPBC2007_GRNs.pdf}, url = {http://dx.doi.org/10.1142/9781860947995_0027} }