A statistical framework for differential network analysis from microarray data

摘要: It has been long well known that genes do not act alone; rather groups of in consort during a biological process. Consequently, the expression levels are dependent on each other. Experimental techniques to detect such interacting pairs have place for quite some time. With advent microarray technology, newer computational interaction or association between gene expressions being proposed which lead an network. While most analyses look differentially expressed, it is potentially greater significance identify how entire network structures change two more settings, say normal versus diseased cell types. We provide recipe conducting differential analysis networks constructed from data under experimental settings. At core our approach lies connectivity score represents strength genetic genes. use this propose formal statistical tests following queries: (i) whether overall modular different, (ii) particular set "interesting genes" changed networks, and (iii) given single networks. A number examples provided. carried out method types simulated data: Gaussian based equations. show that, appropriate choices scores tuning parameters, works data. also analyze real involving heavy mice interesting may play key roles obesity. Examining changes structure can valuable information about underlying biochemical pathways. Differential with useful tool exploring different conditions. An R package be downloaded supplementary website http://www.somnathdatta.org/Supp/DNA .