Modelling transcriptional regulation with fractional order differential equation using Gaussian Process

A reasonable math model is fundamental to describing gene regulatory mechanism. A new method is proposed to model transcriptional regulation with transcription factor from gene expression profiles using fractional order differential equations. Gaussian Process is employed as a tool to model the latent transcription factor activity and particle swarm optimization algorithm is utilized to optimize the fractional order, kinetic parameters in the model and hyperparameters in kernel function. The results of the experiment on real gene expression profiles indicate that the fractional order differential equation fits data better, also the proposed approach is feasible to model transcriptional regulation.