Waste Not, Want Not

When it comes to human waste products, such as urine and feces, the first thing that comes to mind is probably rapid disposal. However, over the last decade, intensive research into their metabolic composition has shown that these waste products harbor a great deal of information relating to diet, lifestyle and risk of disease. Spectroscopic profiling of these samples, typically using mass spectrometry (MS) or nuclear magnetic resonance (NMR) spectroscopy, generates ‘metabolic fingerprints’ or ‘metabolic phenotypes’ that are unique to the person producing them. The technology can characterize a wide range of diseases, as well as being able to track an individual’s response to dietary and therapeutic interventions (1)(2). 

The success of this technology has relied on advances in analytical spectroscopy and progress in bioinformatics. However, there has been much effort but relatively little progress in standardizing methodological profiles, mathematical modeling tools and spectral databases. Some headway has been made in collating databases of reference standards, with the Human Metabolome Database (HMDB; funded by the Canadian Government) and the Biological Magnetic Resonance Bank (BMRB; hosted by the University of Wisconsin at Madison, USA) being amongst the best-known examples. In terms of sharing NMR and MS spectra of real biological samples, there are no large publically-available resources. The European Molecular Biology Laboratory-European Bioinformatics Institute (EMBL-EBI; Wellcome Genome Campus, UK) is co-ordinating an initiative to allow storage, exchange, comparison and re-utilization of metabolomics data and has begun to develop a repository for biofluid spectral data (3).

Yet despite these efforts and the drive towards open-source data by funding and regulatory bodies, investigators generally remain reluctant to deposit ‘their’ data into open-access databases. Combined with variability in analytical protocols, this has severely constrained the utility of the resource. The extent of collaboration, including the practice of data and resource sharing, has reached different levels across academic disciplines. For instance, the proteomic community has a universal standard for reporting data (minimum information about a microarray experiment; MIAME), a plethora of non-commercial freeware (4) and is beginning to provide test-set data with which to validate new algorithms. In principle, the lack of communal facilities does not harm the larger research groups that have sufficient resource to build large in-house databases and data mining tools. However, collaboration across research communities has obvious advantages – the most relevant being to accelerate the pace of discovery.

Barriers to widespread implementation of metabolic profiling technology include the initial capital cost (often prohibitively high for resource-limited research and clinical environments), a lack of trained researchers, a lack of universal standards and annotation, and restricted mobility of both commodities (samples, solvents, equipment) and people. Nevertheless, one of the main attractions of implementing multipurpose metabolic profiling platforms in resource-limited research settings is the conveyance of research flexibility. By focusing a research community or field on critical issues (such as harmonization of experimental design, sample handling, sample acquisition) and by implementing a uniform structure for training and technology transfer, we can begin to both support and benefit from resource-limited research groups on a global scale. The other major initiative required to facilitate success is the sharing of spectral repositories through an accessible framework of databases. 

The efforts in technology standardization to date have been driven by analytical criteria. For smaller research groups the imperative to justify expensive technology with practical application is stronger. Initiatives such as the Latin American Metabolic Profiling Society (LAMPS), which brought together academic spectroscopists from more than 10 countries, supported by industrial partners, such as Airbus, Bruker Biospin, Waters, Agilent and Metabometrix, have arisen in direct response to the need to accelerate research by building networks of chemists and application scientists. The goal is to create active research networks that can foster and deliver collaborations across local research groups in biomedical sciences. 

In addition to the chemical requirements for conducting high-quality research, the group has also identified biological areas of interest across the community, including bariatric surgery and dengue infection, whereby the community can reach a critical mass of data more quickly and has an inbuilt structure for validation of disease-associated biomarkers and metabolic networks. Rather than constraining metabolic phenotyping to just a few of the worlds best-equipped chemistry laboratories, we should work towards reducing the ‘waste’ of locally owned and hidden spectral resources (by identifying opportunities for sharing of facilities, mobilizing scientists and consumables, organizing training and technology transfer networks and creating annotated data repositories and reference databases). Overcoming these barriers will help fulfill the potential of metabolic profiling as a mainstream tool in terms of diagnostics, prognostics and monitoring in both patient- and population-centered frameworks, thereby building a bridge between analytical scientists and clinicians to drive translational healthcare initiatives.