Inference of gene regulatory network using modified genetic algorithm

The major challenge of inferring genetic network is mining the dependencies and regulating relationship among genes. The paper tries to address this problem using Genetic Algorithms to infer the transcription regulatory network. While Genetic Algorithms(GA) are able to infer smaller networks with good sensitivity and precision, several generations and much greater computation power are required to infer regulatory networks from realistic data. Here a modified GA that uses statistical techniques to narrow the search space is proposed. The system is tested on the publicly available datasets of the Hela cell cycle and Yeast cell cycle. The results have been compared with regulatory networks inferred by using second order differential equations. It is found that the sensitivity and specificity are at par with differential equation method and has a considerable improvement in comparison with the Basic GA method.