Main content area

Independent or integrative processing approach of metabolite datasets from different biospecimens potentially affects metabolic pathway recognition in metabolomics

Zhou, Li, Xu, Jin-Di, Zhou, Shan-Shan, Zhu, He, Kong, Ming, Shen, Hong, Zou, Ye-Ting, Cong, Long-Jie, Xu, Jun, Li, Song-Lin
Journal of chromatography 2019 v.1587 pp. 146-154
anemia, biochemical pathways, chromatography, computer software, data collection, herbal medicines, metabolites, metabolomics, models, urine
In metabolomics studies, metabolic pathway recognition (MPR) is performed by software tools to screen out the significant pathways disturbed by diseases or reinstated by drugs. To achieve MPR, the significantly changed metabolites determined in different biospecimens (e.g. plasma and urine) are analyzed either independently (metabolites from each biospecimen as a dataset) or integratively (metabolites from all biospecimens as a dataset). However, whether the choice of these two processing approaches affects the results of MPR remains unknown. In this study, this issue was addressed by selecting evaluation of the effects of the herbal medicine Rehmanniae Radix (RR) on anemia and adrenal fatigue by UPLC-QTOF-MS/MS-based metabolomics as an example. The significant pathways disturbed by the modeling of anemia and adrenal fatigue and those reinstated by treatments with raw and processed RR were recognized using MetPA software tool (MetaboAnalyst 3.0), and compared by independent and integrative processing of the significantly changed metabolites determined in plasma and urine. The results showed that the two processing approaches could yield different impact values of pathways and thereby recognize different significant pathways. The differences appear to happen more easily when metabolites from different biospecimens shared the same metabolic pathway. Such pathway could be recognized as a significant pathway by integrative processing but could be excluded by independent processing due to the converged and dispersed importance contributions of the involved metabolites to MPR in the two processing approaches. This issue should concern researchers because MPR is crucial not only to understanding metabolomics data but also to guiding subsequent mechanistic research.