Mapping metabolomics data: Complexity and issues
Résumé
To optimize the translation of large-scale metabolomics by defining meaningful results, data contextualization is mandatory. Although a number of tools and methods have been developed, there is still no standardization of practices.
In this context, the objective of the work was to evaluate pathway analysis to biologically contextualize metabolomics data, identify bottlenecks and optimize workflows to provide reproducible information able to guide biological interpretation. To fulfil this objective, a published dataset including a list of identified metabolites modulated with metabolic syndrome in elderly men was used (Comte et al. 2021).
A large number of tools using neither the same methods nor the same databases were first evaluated. Then, four alternative mapping tools (i.e. ConsensusPathDB, MetaboAnalyst, MetExplore and RaMP) enabling to cover a wide range of methods, from the use of a single metabolic network to the use of multiple databases with different identifiers were deeper investigated.
Our work showed that, before taking an interest in mapping methods, it is essential to produce complete lists of adequate identifiers regarding the used databases or network, which often results in loss of information. Using multiple pathway databases was found to be a good strategy to derive a consensus pathway signature and increase the metabolome coverage. For fully identified metabolites, the use of metabolic network and subnetwork extraction appeared to be more pertinent to go deeper into metabolic exploration. In each case, the level of knowledge about the annotated metabolites, as well as the contextualization objective should guide the design of optimal workflows.
Origine | Fichiers produits par l'(les) auteur(s) |
---|