Complex diseases involve extensive metabolic interactions within intricate biological networks. Consequently, it is advantageous to analyze metabolic phenotype data through metabolite interactions rather than focus on individual metabolites in isolation. In this article, we propose a novel analysis strategy called SynNet, which constructs multiscale synergy networks associated with specific metabolic phenotypes, offering new perspectives on the metabolic response mechanisms of diseases, including the mechanisms underlying disease comorbidity. The SynNet strategy begins with the construction of a metabolite-level synergy network (m-SynNet). This network is based on the definition and identification of significant metabolite pair interactions that distinguish disease phenotypes. Subsequently, a pathway synergy effect is defined by mapping these synergistic metabolite pairs onto the predefined metabolic pathways and performing a hypergeometric test to assess the probability of these pairs affecting a given pathway pair. The resulting significant pathway pairs identified form a pathway-level synergy network (p-SynNet). Both m-SynNet and p-SynNet offer complementary insights into disease mechanisms that go beyond conventional metabolomics analysis. For example, nodes with high connectivity in m-/p-SynNet suggest a strong correlation with the phenotype, while shared pathways across different phenotypes offer clues about the mechanisms of disease comorbidity. We applied the SynNet strategy to two real-world metabolomic data sets of disease comorbidity and identified key pathways associated with disease comorbidity from the p-SynNet. The candidate pathways are supported by the existing literature. Thus, the SynNet strategy may represent an alternative approach for metabolomic data analysis, providing novel insights into disease mechanisms and comorbidity.