Dysregulation of plasma lipids, especially cholesterol levels, has been identified as a critical risk factor for various complex human diseases such as cardiovascular diseases and Alzheimer's disease. Recent genome wide association studies (GWAS) have implicated a number of susceptibility loci for complex traits of lipid metabolism. However, these loci only explain a limited proportion (25-30%) of the genetic heritability and a comprehensive understanding of the underpinning regulatory mechanisms has not been achieved. To better understand the mechanisms and to identify potential novel regulators, in this study, we utilized an integrative genomics approach that leveraged multiple genetic and genomic datasets including 1) GWAS from Global Lipids Genetics Consortium (GLGC) for four lipid traits (high density lipoprotein cholesterol [HDLC], low density lipoprotein cholesterol [LDLC], triglycerides [TG], and total cholesterol [TC]), 2) expression quantitative trait loci(eQTLs) from genetics of gene expression studies of human tissues related to lipid metabolism(such as liver and adipose tissues), 3) knowledge-driven biological pathways, and 4) data-driven regulatory gene networks. The integration of these diverse data sources enabled tissue-specific investigations on whether the genetic variants associated with lipid traits from GWAS were concentrated on specific parts of gene regulatory networks, termed "subnetworks", and whether novel key regulators of the lipid subnetworks could be identified based on data-driven network structures. We identified 17, 16, 18, and 14 subnetworks for HDLC, LDLC, TC, and TG, respectively. Among these, subnetworks associated with lipid metabolism, protein modification, response to external stimulus, receptor activity, and neurophysiological process were shared among the four lipid traits. Among the trait-specific subnetworks, those involved in glutathione metabolism, and taurine and hypotaurine metabolism were found to be HDLC-specific while cadherin-associated subnetworks were LDLC-specific. Finally, by utilizing the gene-gene relationships revealed by the network architecture, we detected key regulator genes, both known (e.g.APOE, APOA5,and ABCA1) and novel (e.g.NCAM1, and F2), in these lipid-associated subnetworks. Our results shed lights on the complex mechanisms underlying lipid metabolism and highlight potential novel targets for lipid regulation. [ABSTRACT FROM AUTHOR]