Predicting Single Genes Related to Immune-Relevant Processes

In this paper we address the problem of predicting gene activities by finding gene regulatory dependencies in experimental DNA microarray data. Only few approaches to infer the dependencies of complete gene interconnectivity networks can be found in the literature. Due to the limited number of available data, the inferring problem is under-determined and ambiguous. Therefore, we introduce a new algorithm to infer relationships only between selected genes and the unknown gene network. This method is able to predict gene activation by mathematical modeling of the network and its simulation. The parameters of the mathematical model are determined by optimization with evolutionary algorithms. In this paper we will show that our approach is able to correctly predict gene responses in immune related regulatory processes and correctly identify some of the true genomic relationships of these genes.