Introducing ACO+ for Modeling Gene Regulatory Networks from Microarray Data

Revealing the fundamental cellular progression through systems biology is a long felt task and application of computational models in Inferring Regulations among the genes is a suggested approach to be optimized. Various models have been used to extract the probable structure and dynamics of such networks from gene expression data. However, capturing the complex nonlinear system dynamics is still a big issue to be worked out. In this paper a method based on a unique algorithm named ACO+ has been proposed for reverse engineering Gene Regulatory Network (GRN) from microarray data using the biologically relevant optimization algorithm Ant Colony Optimization (ACO), with an enhancement by incorporating two genetic operators, crossover and mutation. The Linear Time Variant (LTV) Model, being simple and able to simulate also nonlinear system dynamics, has been used for modeling the GRN in a fitted way. Attempting ACO+ to model GRN is of its first kind and optimization has been carried out using ACO and the variants ACODE (ACO followed by Differential Evolution), and ACO+ with synthetic noise free and noise in data. The applicability of the proposed method has been tested on synthetic datasets as well as real expression data set of SOS DNA repair system in Escherichia coli. Of the algorithms attempted, ACO+ simulated the best optimization.