Demonstrating that the data produced in metabolic phenotyping investigations (metabolomics/metabonomics) is of good quality is increasingly seen as a key factor in gaining acceptance for the results of such studies. The use of established quality control (QC) protocols, including appropriate QC samples, is an important and evolving aspect of this process. However, inadequate or incorrect reporting of the QA/QC procedures followed in the study may lead to misinterpretation or overemphasis of the findings and prevent future metanalysis of the body of work.
The aim of this guidance is to provide researchers with a framework that encourages them to describe quality assessment and quality control procedures and outcomes in mass spectrometry and nuclear magnetic resonance spectroscopy-based methods in untargeted metabolomics, with a focus on reporting on QC samples in sufficient detail for them to be understood, trusted and replicated. There is no intent to be proscriptive with regard to analytical best practices; rather, guidance for reporting QA/QC procedures is suggested. A template that can be completed as studies progress to ensure that relevant data is collected, and further documents, are provided as on-line resources.
Key reporting practices
Multiple topics should be considered when reporting QA/QC protocols and outcomes for metabolic phenotyping data. Coverage should include the role(s), sources, types, preparation and uses of the QC materials and samples generally employed in the generation of metabolomic data. Details such as sample matrices and sample preparation, the use of test mixtures and system suitability tests, blanks and technique-specific factors are considered and methods for reporting are discussed, including the importance of reporting the acceptance criteria for the QCs. To this end, the reporting of the QC samples and results are considered at two levels of detail: “minimal” and “best reporting practice” levels.
Progress in science is based on the principle that measurements are repeatable; that is to say, it should be possible for scientists in other laboratories, equipped with the same, or similar expertise, resources and infrastructure, to replicate the results of published work within acceptable levels of confidence. Properly documented studies allow other scientists to critically assess and judge the original experimental design and should also enable researchers in other laboratories to repeat the experiment under the same conditions and support or refute the findings. However, a reproducibility crisis in science has been recently highlighted by Munafò et al. (2017) and the lack of transparency in reporting can lead to ambiguity within experimental design, data collection, as well as data processing and interpretation, a situation which is also true for metabolomics. Moreover, any metabolomic data derived from instrumental analyses must be sufficiently reliable that decisions based on it can be reported with confidence; there should be no “leap of faith” from data to knowledge. If the metabolomic data cannot be trusted, then the answer to the posed question is of little value.
Discussion and conclusions
The use of both QA and QC procedures is recognized as fundamental for achieving reliable and consistent data in metabolomics. The recommendations presented here are not meant to constrain research but to ensure that, where these systems have been employed, this is stated and what has been done is clearly explained. In this opinion paper there is no official authority to mandate the procedures that must be used when employing QA/QC in untargeted metabolomics. However, the recommendation here is that authors state what they did, and to state this as transparently and as completely as possible. In other words, they should deliberately and purposefully avoid obfuscation. The documentation of the QA and QC of the analytical process of an untargeted metabolomics experiment enables experimenters to demonstrate that the work was conducted appropriately, and that the editors and reviewers, and therefore ultimately readers can have confidence in the resulting conclusions. However, recognizing that untargeted metabolic phenotyping is a discovery methodology any hypotheses emerging from it are inevitably provisional. Nevertheless, armed with knowledge that the data are of good quality provides the motivation to devote the resources required to undertake further targeted analysis, using quantitative and validated methods, to confirm, or refute, these hypotheses.