Reconstructing gene regulatory networks by integrating knockout and multifactorial perturbation data

Structure identification is investigated in this paper for a gene regulatory network (GRN) using two different types of steady state expression data. Using knockout expression data, direct regulating genes for an arbitrary gene are estimated by means of maximum likelihood estimation (MLE). And, using multifactorial perturbation data, a weight corresponding to a possible direct regulation is constructed. To take above information into consideration, the weight matrix is further adjusted based on these estimated direct regulatory relationships. Computation results with the DREAM4 In Silico data show that prediction performances are substantially improved by the suggested method, which reveals that information contained in both knockout data and multifactorial data is complementary. Furthermore, the high precision of the obtained most reliable predictions shows the suggested algorithm may be helpful in guiding biological experiment designs.