Lipidomics is of growing importance for clinical and biomedical research due to an increasing number of discovered associations between lipid metabolism and diseases. However, sophisticated computational methods are required for biological interpretation including an understanding of metabolic processes, and their underlying (patho)mechanisms from lipidomics data. This can be achieved by using metabolic networks in combination with graph algorithms. Here, we present a lipid network analysis framework (Lipid Network Explorer, short LINEX) that allows biological interpretation of changes in lipidome composition. We developed an algorithm to generate data-specific lipid species networks from reaction database information. Using these networks, we developed a network enrichment algorithm that infers changes in enzymatic activity from lipidomics data, by leveraging multispecificity of lipid enzymes. Our inference method successfully recovered the MBOAT7 enzyme from knock-out lipidomics data. Additionally, we predict a PLA2 member to be at the center of dysregulation in adipose tissue in obesity. In our work, we showed the potential of lipidomics data to unravel mechanisms of metabolic regulation. Thereby our presented method can make lipidomics more clinically relevant by elucidating potential disease mechanisms.