Professor Christoph Steinbeck discusses PhenoMeNal – a new online portal that allows researchers and clinicians to perform large scale metabolomics data analysis.
An international collaboration between EMBL’s European Bioinformatics Institute (EMBL-EBI) and 13 other partners has made large-scale metabolomics analyses easier with the launch of PhenoMeNal.
Metabolomics is the study of small molecules that exist in cells, biofluids, tissues and organs. These molecules (metabolites) and their concentrations reflect the underlying biochemical activity and thereby the health of cells or tissues. Metabolomics could become an essential tool for precision medicine, helping with patient diagnosis and possible treatment options.
PhenoMeNal allows researchers to perform such highly complex metabolomics analyses. For example, researchers can search for patterns in a patient’s data, and use the findings to improve the detection of disease and to help optimise treatment.
Speaking to SciTech, Professor Christoph Steinbeck, scientific co-ordinator of PhenoMeNal, explains why the online portal is important, what it can achieve, and some of the challenges involved in its development and roll-out.
Why is a tool such as PhenoMeNal important to the research community? What are its biggest benefits with regard to the future of precision medicine?
The most important objective for an infrastructure such as PhenoMeNal is to lower the entry barrier for researchers (and that includes clinical researchers on the computational side), so that they can access computational metabolomics workflows more easily. Before we created PhenoMeNal, the landscape essentially consisted of individual laboratories would take diverse sets of existing metabolomics tools and try to integrate them into those workflows that worked for them. We have introduced a huge change, here.
Regarding infrastructure, it is, of course, one of the aims of the European Commission – via its infrastructure calls – to provide best practice tools that are used by a variety of people so that standard operating procedures and protocols can be implemented as reusable work flows, and PhenoMeNal certainly ties into that. Indeed, through the project we are providing an Infrastructure as a Service (IaaS) for computational researchers in the clinic, and if people use this infrastructure and feedback standard operating procedures and protocols into the infrastructure, then they become immediately available for other people working in the field.
The main aim of this infrastructure was to benefit the clinical research space, but we can also offer services to any other complex type of research where it is necessary to have a water-tight audit trail of what actually happens to the data while it is being processed.
Regarding the benefits for precision medicine, this, first of all, comes back to lowering entrance barrier for clinical scientists to actually work with metabolomics data. Metabolomics has fantastic potential to allow medicine to move from a single biomarker to more complex patterns, using the whole metabolome as a biomarker. But to really understand how all the different influences acting on a patient modify the metabolome – and there are numerous such influences, including exercise, nutrition, medication, and so on – is a big challenge. This is essentially a big data problem where data and meta-data from many patients need to be linked to their metabolome measurements and the changes therein. PhenoMeNal can help here by providing best practices on how to modify and work with this data.
With infrastructures like PhenoMeNal we are also moving towards what is termed ‘research objects’, which are a combination of the clinical data that people are working on with the protocols, the parameters, and the best practices, all implemented in executable work flows. And this research object then enables researchers to replicate, for instance, an entire publication in the scientific literature at the click of a button because not only is the data there but so are the executable work flows.
Our colleagues at Uppsala Medical Hospital, Sweden, for example, have recently used metabolomics data as biomarkers to diagnose secondary progressive multiple sclerosis very early in the disease progression, which can be held up as an impressive demonstration of the usefulness of an infrastructure such as this.
What were the biggest challenges involved in creating the online portal? How important was it for the EMBL-EBI to work with external partners?
We experienced challenges on both the usability and technical sides. We wanted to build a cloud-based infrastructure so as to ensure that PhenoMeNal is as portable as possible. But building cloud services means that you are always aiming at a fast-moving target; if you want to build things on the latest technology, which was of course what we wanted to do, then you have to keep up with some very big industry players – such as Amazon, Google, and Microsoft – who are moving forwards with their technologies at an incredible rate. As such, we were constantly challenged here, but we were nevertheless able to develop and through consolidation efforts we have been able to keep up.
Integration into the existing European Commission-funded infrastructures was also key – there are already cloud storage and cloud computing infrastructures up and running which we wanted to collaborate with – and this also presented certain challenges but we eventually succeeded.
In terms of usability, the biggest challenges concerned improving user experience; if you are a technical expert or an informatician, then you may build things in a way that might not necessarily be what your users would like to see. At the EMBL-EBI, some years ago they moved to user experience as a basis for creating their infrastructures, and we were able to build on this experience. We also hired user experience experts to work with our user community and so improve the user interface and documentation. And, last but not least, building excellent documentation that enables people to understand the complex infrastructure was also crucial, but we eventually rose to the challenge.
How have you worked to ensure that patient data privacy and ethical considerations are adequately taken into account?
It was key to work with domain experts in those fields, and so our main approach was to appoint an ethics advisor, who was already a domain expert in the field, and whom we regularly approached for advice. We also ran regular workshops for ourselves where we invited a number of ethics and privacy experts in the medical field to improve our own understanding of this area.
This nevertheless remains a challenge because of the multitude of different regulations that exist in the different European EU countries – which are in addition to the over-arching European law – and because of that, and because of the extreme complexity of the field, we eventually decided to take a very conservative solution, in that we keeping our patient data behind the firewalls of hospitals and research institutions who have the approval to work with that type of data, and we make it absolutely clear to our users that they must not use un-anonymised data in the public-facing demonstration part of the infrastructure.
Do you feel that there are any barriers to the wider use of the portal by researchers? If so, how do you hope they can be overcome?
There are perhaps multiple components to this. PhenoMeNal was developed over a three year period, and while we have come a long way it nevertheless remains a technical infrastructure with a learning curve.
The work flows and infrastructures we are providing, along with the documentation, make it now much easier to use, but it is still important for people to understand that this is not a one click solution. You still have to understand what you are doing, and you have to know your parameters, which is why we have exposed all of these for all of our components. It can still take a while for people to use the portal due to the technical dimension of some of the barriers, and so we will continue to work in the coming years to lower those barriers even further.
What might be termed the ‘attention economy’ can also present a barrier, in that it is increasingly difficult in the modern world to catch the attention of the people who might use your product. Indeed, publicising our existence is a challenge, and so we are writing articles, and we attend conferences where we speak about the infrastructure, but that is not necessarily enough; it requires really a concerted effort on the part of the entire consortium to continue publicising the infrastructure.
Moving forwards, what are your hopes for the future of the portal? How do you hope to maintain and develop the portal in the future?
We had a fantastic and enthusiastic consortium with over 60 scientists and engineers working for three years to make PhenoMeNal happen, and the great thing is that for a large proportion of our collaborators, including myself, PhenoMeNal is helping to solve some of our own problems, and that is a key component to create a sustainable and successful project that continues into the future.
We will all continue to use our own infrastructure for our own research, but of course there is also the need to formalise the sustainability efforts, and to this end we have worked with various communities over the years, the most important being the ELIXIR infrastructure for life science in Europe, where we have been successful in creating the ELIXIR metabolomics community which will help keeping the consortium together. And this will not only consist of the PhenoMeNal consortium; it is a community of metabolomics researchers, from both wet and dry labs, in Europe.
Equally important is the fact that we have recently founded the European Metabolomics Infrastructure Foundation. The main aim here, as the name implies, is to support and develop metabolomics infrastructures in Europe, and that is not limited to e-infrastructures like PhenoMeNal, although the portal will be a key target; we are also aiming to build metabolomics wet lab infrastructures where scientists can run their experiments and get advice on best practices.
We have also been feeding a lot of components back into the various communities from which we have actually benefitted, such the Galaxy community which was building the fantastic workflow tool that we have been working
with. Feeding back into those communities ensures that the components that we are developing are maintained, stay alive, and are continuously updated.
How will the Foundation complement the work that is taking place at the EMBL-EBI?
On the infrastructure side, EMBL-EBI has a particular mission, which is to build data infrastructures for biomedical data; but it is just one player in a whole set of localised infrastructures across Europe which all play a role in metabolomics, and there is still much more to do.
ELIXIR was created realising that EMBL-EBI cannot and should not do everything in biomedical infrastructures, but that national nodes and networks of nodes in the member countries must play a vital role too. The Metabolomics community, where of course EMBL-EBI plays a major role either, is such a network.
The aim of the Foundation is to create a legal entity that brings the relevant people together in one place and so provide a not for profit body that supports the further creation of metabolomics infrastructure in Europe.
Professor Christoph Steinbeck
Institute for Inorganic and Analytical Chemistry
+49 3641 948171