Modeling the temporal evolution of the Drosophila gene expression

摘要

The evolution of the gene expression pattern of Drosophila, from the embryonic to adult development phases, was studied on the basis of a microarray time series involving the expression levels of 4028 genes over 67 time-points. The genes presenting a similar temporal evolution of their expression levels were clustered together, so as to define a small number of representative classes. To model the network interactions responsible for the dynamic behavior of gene expression, a system of linear differential equations with constant coefficients was used. The parametric estimation of this model was performed in two stages: a first stage of linear parameter identification allowing an analytical approach to the solution, and a second stage of nonlinear parametric estimation which refines this solution. This model is shown to reproduce the experimental gene expression profiles with a fairly good precision.