Abstract
How to construct gene regulatory networks from microarray data has attracted research attention in recent years. Time-series profiles of gene expression generated by DNA microarrays possess rich information to construct dynamic models of transcription behaviors. In this study, with the help of correlation clustering, the AutoRegressive with eXogenous input (ARX) models is introduced to construct a gene regulatory network in response to heat shock in the environment and may provide new insights into the thermo-tolerance mechanism of biological processes under heat shock stress. If all microarray data and the related genes are included, the gene regulatory network would become too complex to get their important regulatory relationships. So a cluster regulatory network is constructed at first to simplify the construction procedure of the complex gene regulatory network of heat shock. Therefore, the fuzzy K-means algorithm is used to cluster the genes with similar expression profiles under heat shock stress. Then a cluster gene regulatory network for heat shock is constructed based on the ARX model and the averaging method. The sparseness of the cluster gene regulatory network is also considered using Akaike’s Information criterion. Finally, we construct the gene regulatory network of heat shock based on the interactive information of the constructed cluster regulatory networks to find some possible gene regulatory mechanisms under heat shock stress.