Inference of genetic regulatory networks from time-series gene expression data has gained significant attention in the bioinformatics community over the last decade. It has been shown to be impractical to infer these networks by devising gene disruption experiments on a genomic scale. Various computational models and inference algorithms have been proposed to overcome this problem. The most prominent among these is the REVEAL algorithm which uses an information theoretic approach. The REVEAL algorithm faces a potential problem of identifying indirect regulatory relations between genes. In this paper, we propose a modification to the REVEAL algorithm along with a graph theoretic algorithm to eliminate the indirect regulatory relations. Our algorithm is not just limited to REVEAL algorithm, but can be applied to any algorithm which infers indirect regulatory interactions. Our algorithm has been tested and compared with the REVEAL algorithm on synthetic data. Thus, we were able to infer direct and more meaningful regulatory relations between genes efficiently.
展开▼