Multi-level Regulation Of Gene Expression

Regulatiuon of eukaryotic gene expression is a complex process,including transcriptional,post transcriptional,translational,posttranslational regulation,etc.Dynamical and accurate regulation of these process play a vital role on the efficiency and accuracy of gene expression.Recently,more and more researchers place high hopes on the application of gene expression regulation in various disease diagnoses and therapies.The identification of multi-dimensional gene regulatory modules is deeply explored in this paper,which based on Sparse Multi-Block Partial Least Squares and Particle Swarm Optimization.Firstly,this paper uses the Sparse Multi-Block Partial Least Squares to identify multi-dimensional regulatory modules.We demonstrated the performance of our method on the TCGA Ovarian Cancer datasets including the copy number variation,DNA Methylation,microRNA expression and gene expression data.Through the program we obtains 34 regulation modules,and then we describes the biological significance of these 34 modules.At the same time obtain the heat map for feature profiles of copy number variation,DNA methylation,microRNA and gene expression in modules.Secondly,due to Sparse Multi-Block Partial Least Squares exhibits subjectivity for the selection of parameters during the analysis of the modules,on the base of Sparse Multi-Block Partial Least Squares,we introduced Particle Swarm Optimization,which is used to optimize the objective function parameters and makes CV score of cross validation as the fitness value.Through verification,we can obtain more optimal results by selected proper parameters,which makes the results more statistically significant,and has a better practical guiding significance.