Research On Reverse Engineering Of Gene Networks With Regulation Theory

DNA microarray is a new technology with the development of "Human Genome Project". High-density DNA microarrays load thousands of DNA fragments that can be used to detect high-throughput biology, computer data processing and information mining of which are the focus of recent research. Some first aid of statistical methods for gene clustering which is the initial screening of gene expression data, and further study of gene interaction networks(namely reverse engineering), have become an important research content of systems biology. Prediction of gene regulatory networks has two steps:First, clustering methods can be used to divide the massive gene expression data into dozens or hundreds of small-scale gene classes, and then we can build gene regulatory networks in a small area through reverse engineering methods.Here we propose that L1/2regularization can be used to reverse engineer a gene regulatory network from time-series data, combines ordinary differential equations, sigmoid function and cross-validation for determining the in-degrees, and get a sparse network structure. The result of the algorithm is an undirected network, in which each edge has been assigned a score from a bootstrap procedure. P-R and ROC theories are used to evaluate the performance.