6. How do you see the field evolving over the next 5–10 years? What are the main obstacles to be overcome?

Dajana Vuckovic: “I believe that in the next 5–10 years we will see additional instances of successful validation and translation of metabolomics results into the clinic. There will also be more and better integration of temporal and spatial information into metabolomics studies through increased use of fluxomics and MS imaging, respectively. In addition, integration of metabolomics with other omics will continue to advance, and provide us with new biological insights and new pathways towards personalized medicine.

One of the major obstacles we need to overcome for untargeted metabolomics is to increase metabolite coverage. Many of our standard workflows do not let us detect metabolites present in low concentrations, which are often important in biological signalling and pathway regulation. Together with strategies to address matrix effects and metabolite identification bottleneck, better metabolite coverage will allow us to build a more accurate picture of metabolite networks.

A second grand challenge is to collect and add additional biologically relevant information into our existing metabolite databases: circadian variability, inter-individual variability, intra-individual variability, stability of metabolite in a given biological sample, extent of binding of a given metabolite in plasma or tissue, etc.  Such information is critical to improve our biomarker discovery efforts and help us rationally design metabolomics studies with sufficient statistical power for a given application.”

Ian Wilson: “The analytical methodologies are becoming more robust and their limitations better understood. I see the field evolving into faster methods of metabotyping and biomarker discovery. Increasingly, as we begin to better map the metabolome, we will move from untargeted profiling where the majority of the metabolites are uncharacterized at the beginning of the study and unidentified at the end to a situation where profiles are fully annotated.

What are the main obstacles to be overcome? To be honest, I see no technical/informatics obstacles that cannot be overcome with time and the application of intelligence and resource.”

Fengguo Xu: “Metabolomics is moving to large-scale targeted metabolomics from an untargeted one. In the past few years, more attention was paid to disturbed metabolites profiling induced by diseases, drugs and environment. Increasing key differential metabolites were screened out. Unfortunately, the differences in sample preparation, data pre-processing and multivariate data analysis may lead to varied interpretation. The inter-laboratory reproducibility of untargeted metabolomics was confusing. Thus a follow-up targeted metabolomics study should be conducted to validate the untargeted metabolomics findings. Although data obtained from targeted metabolomics platforms are relatively reliable and reproducible. But its application was limited by its low coverage of detectable metabolites.

As described above, the resulting differential metabolites screened out by untargeted metabolomics methods are influenced by varied data pre-processing. One of pitfalls is how to make the datasets more accurate before data analysis. Metabolomics has faced big data challenges, including data standardization, management and integration. A guideline for regulating this key section of metabolomics should be on the table as soon as possible.

Besides, the interpretation for biological functions of key differential metabolites and corresponding pathways is becoming a key driver for clinical and diagnostic purposes.”