Using Gaussian Graphical Model(GGM) to find time-series networks rewiring. BrainSpan time periods data.
Inferring a network which "well" modeling chmorosome modification genes and RNA binding proteins in MICRF models.
Three main references:
-
inferring dynamic gene regulatory networks in cardiac differentiation through the integration of multi-dimensional data, --> main idea to penalty two continue time points.
-
Sharing and Specificity of Co-expression Networks across 35 Human Tissues, --> main idea to Gaussian Graphical Models and additional penalty.
-
FastGGM, main idea to give lambda and fast running
Tools for solving the GGM models: convex optimization model: 1, use the similar idea with GNAT (ref.2). 2, use the matlab tools CVX http://cvxr.com/cvx/. (ref.1).
Prior information or result validation: 1, based on all the expression data to infer one density network as a prior network, or prior given zeros edges which would be decreasing the computational complexity. 2, use well-build database as prior networks, such as, COEXPRESdb and TS-CoExp
Main idea: see report 11_3 and 10_27 slides