A Network-Based Approach to Analyze and Integrate Untargeted Metabolomics with Various High-Throughput Experimental Data

Integrative analysis of metabolites is essential to obtain a comprehensive view of dysregulated biological pathways leading to a disease. Despite the great potential of metabolites their system level analysis has been limited. Global measurements of the metabolites by liquid chromatography-mass spectrometry (MS) detects metabolites features changing in a disease. However, identification of each feature is a bottleneck in metabolomics, in which a fraction of them are identified via tandem MS. Consequently, the scarcity of these data add additional barriers to decipher their biological meaning, especially in relation to other 'omic data such as proteomics. To address these challenges, a novel network-based approach called PIUMet is described. PIUMet infers dysregulated pathways and components from the differential metabolite features between control and disease systems without the need for the prior identification. The application of PIUMet is demonstrated by integrative analysis of untargeted lipid profiling data of a cell line model of Huntington's disease. The results show that PIUMet inferred dysregulation of sphingolipid metabolism in the disease cells. Additionally, PIUMet identified disease-modifying metabolite in the pathway that remained undetected experimentally. Furthermore, the lipidomic data of these cell lines was integrated with global phospho-proteomic ones. Integrative analysis of these data using PIUMet was shown to systematically lead to identifying dysregulated proteins in the disease cells that cannot be distinguished with individual analysis of each dataset.