Fitness function for evolutionary system to predict unknown gene regulatory networks

This paper proposes a method, denoted pulse flux analysis, to evaluate computationally generated gene regulatory networks, without using biological knowledge related to the target gene regulatory network. Furthermore, the quantification of networks enables the ranking of predicted gene regulatory networks, and prioritization of network candidates to examine by biological experiments is also possible. A short pulse is injected to input nodes of the target network, and the response behavior is analyzed. The method also detects logical ambiguities that cannot be revealed by methods that analyze the static structure of networks. The presented method is incorporated into our system to predict gene regulatory networks, which relies on evolutionary mechanism and high simulation speed.