Investigating Metabolic Pathways of Ankylosing Spondylitis via Compound Similarity Network-Assisted Metabolomics Analysis.

Abstract

LC-MS-based metabolomics is a powerful tool in analyzing disease molecular mechanisms. Because of its high sensitivity and throughput, LC-MS-based metabolomics usually detects thousands of metabolites. How to find disease-related metabolites and investigate metabolic pathways is critical in metabolomics studies. Conventional statistics-guided data mining looks only for mathematical relations between the detected metabolites and the metadata. It is not enough to unveil biological pathways of metabolites involved in disease progression. Compound similarity network (CSN) is a spectral-independent technique to cluster compounds based on their structural similarities and to investigate potential chemical transformations. Herein, we developed a CSN-assisted metabolic data mining strategy to quantitatively find key metabolites in diseases through structural similarities and explore disease-regulating metabolic pathways based on KEGG and RetroRules metabolic reaction templates. The strategy was used in a metabolomics study of ankylosing spondylitis (AS), in comparison with a healthy cohort and rheumatoid arthritis (RA), a rheumatic disease having similar symptoms with early AS. Using CSN-assisted data mining, a palmitic acid pathway was constructed, which may be regulated in AS pathogenesis.