Roy Goodacre is Professor of Biological Chemistry at the University of Liverpool. At Liverpool he is a director of the Centre for Metabolomics Research, Chair of the Institute of Systems, Molecular and Integrative Biology (ISMIB) research committee, and Deputy Head of the Department of Biochemistry and Systems Biology.
Roy is the founder and current Editor-in-Chief of the peer-reviewed scientific journal Metabolomics. A Director of the Metabolomics Society (2005-2015 & 2020-) and since 2008 also a Director of the Metabolic Profiling Forum.
Roy’s full biography can be found here 
1. When and why did you start using metabolomics in your investigations?
That’s an interesting question, and one that actually has a rather long answer. This is because metabolomics as a discipline didn’t exist when I started conducting research.
I started my PhD in the late 1980s where I was developing and applying pyrolysis-mass spectrometry (Py-MS, not to be confused with Pimm’s…) for the characterisation of bacteria. What I did was genetically modify bacteria by inserting plasmids (and curing some bacteria of these mobile elements), looking at gene knocking outs, as well as altering growth media to elicit specific phenotypic effects. These were then analysed using Py-MS. Pyrolysis was used to get non-volatile material (i.e., bacteria) into the gas phase for MS analysis. Back then MALDI and ESI were only just starting to be developed and applied, and of course people will remember that Koichi Tanaka and John Fenn shared the Nobel Prize in Chemistry in 2002 for developing MALDI and ESI for analysis of biological macromolecules. This was 14 years after my initial Py-MS experiments.
These pyrolysis-mass spectra were complex (150 ion count values at amu from 51-200 m/z) and so I had to learn how to use multivariate analysis (chemometrics) within Genstat on a mainframe computer. That was quite scary – although not as scary as my PhD viva as I stupidly put all the equations for PCA, CVA and hierarchical clustering in my thesis. My internal examiner spent some considerable time asking me to derive these. I still get cold sweats when I think of this!
After my PhD in Bristol, I did a post doc with Doug Kell in Aberystwyth where we developed Py-MS for quantitative analysis of bioprocesses using artificial neural networks (ANNs) and partial least squares regression (PLSR) for analysis of these spectra. During this time I went to many analytical pyrolysis conferences and met up with similar minded scientists. I met Rick Higashi from the University of Kentucky at these and he was using Py-GC-MS to understand soil chemistry. Rick as you know is a fellow ‘metabolomer’ and when we reconnected on the metabolomics conference circuit a decade or so later, we were both bemused and happy to see one another again.
This history tells you that I was already using bioanalytical methods based on mass spectrometry to look at biological systems, and of course to analyse these data using chemometrics. This sounds a lot like metabolomics to me! I'm not the only person to have said this. Jan van der Greef from TNO and Leiden University, who was an early adopter of metabolomics for investigating health and disease, had come across the pioneering early work in Py-MS for bacterial analysis in The Netherlands by Henk Meuzelaar and Jaap Boon. Jan along with Age Smilde wrote about this scientific evolution of metabolomics with respect to chemometrics in 2005 where they mention pyrolysis-mass spectrometry being part of that journey [see J. Chemometrics 19, 376-386: https://doi.org/10.1002/cem.941].
2. What have you been working on recently?
Our research group is quite agnostic in terms of the sorts of biological systems that we analyse. These range from the analysis of human systems in health and disease, mammalian systems, as well as plants and microbial systems – in fact pretty much anything you can squirt into a mass spectrometer!
That said, what currently drives us at the University of Liverpool is to try and understand bacterial responses to stress. We have worked in the past looking at bacterial adaptation when they encounter human drugs – this was initially from an environmental point of view, as many pharmaceuticals end up in the environment. Rivers are full of, amongst other things, anti-inflammatories like ibuprofen and diclofenac, as well as the beta-blocker propranolol, and these drugs may inadvertently alter metabolism within microbial communities. I also wonder what bacteria think of when they get a good glug of caffeine when we’ve excreted this after our morning fix! Do they get a similar buzz? I don’t know.
An environment closer to home is the microflora within your gut. I think that it will be important to understand how this complex community responds to these stresses, especially as the human population is ageing and many people take drugs to stay fit and healthy. What is also becoming increasingly recognised is the link between bacteria and disease where there doesn’t seem to be any obvious microbiological link. The gut microbiome in humans has been implicated in many human diseases including amongst others Alzheimer's and Parkinson’s. A great table summarising these links is in this review by Etheresia Pretorius and Doug Kell [Table 3: FEMS Microbiol. Rev. 39, 567-591: https://doi.org/10.1093/femsre/fuv013].
In addition to working in this area, we do have a particular focus on antimicrobial resistance (AMR). This urgently needs to be addressed and if you’ve read Jim O’Neill report on AMR , he has stated that if left unaddressed AMR bacteria could kill 10 million people per year by 2050, this would surpass the current cancer mortality.
3. What are the main challenges for developing mass spectrometry-based metabolomics for long-term studies?
I think there are several and most of them are not the analytics itself. The challenges are the wrapping at each end of the MS analyses and these are bound in computational analyses in terms of actually planning these experiments and what you do with the metabolomics data once they have been generated.
During my career I’ve had the privilege of working with some outstanding scientists in this area. Most notably Dave Broadhurst and Rick Dunn, where we along with Doug Kell, Ian Wilson and many others established a series of protocols that allowed long-term collection of data over periods of 12 to 18 months. For those who haven’t seen our paper from 2011 it is here [Nature Protocols 6, 1060-1083: https://doi.org/10.1038/nprot.2011.335]. The key thing that we developed was the use of biological QCs that are analysed throughout the GC-MS and LC-MS runs.
These QCs serve two purposes. The first is a rather basic one – if one uses an unsupervised learning method like PCA to analyse all the data (samples plus QCs) then in the PCA scores plots, if the QCs cluster together very close to the origin (they are of course the average sample) then one can have confidence in the reproducibility of the experiment. The second which really Dave pioneered, was to use the QCs to correct for local instrument drift throughout these analytical runs. I think the main challenge is to get this right. And, of course, how to do this when you don’t have the ideal QC at the start of the experiment. This is something we have tried to address in a recent review we published in Metabolomics in 2018 [14: 72: https://doi.org/10.1007/s11306-018-1367-3].
4. You have developed a variety of different Raman spectroscopy approaches for bioanalysis. What do you think is the future of Raman spectroscopy in metabolomics?
I think Raman spectroscopy can be highly beneficial to metabolomics in several areas.
Generally speaking, Raman is not chemically specific and gives an overall biochemical fingerprint of the sample. The provenance of the sample can then be assessed using chemometrics (the same multivariate analysis that people use in MS or NMR metabolomics). The advantage that this method has is that it can be delivered in a handheld portable format. Therefore, the spectrometer can go to the patient rather than the other way round. This could be readily employed in patient triage scenarios. Although not for clinical purposes, I used a handheld Raman instrument to measure Aboriginal Rock art in Kimberly, Australia in 2019 . The aim was to use Raman spectroscopy to investigate the chemistry of pigments used in the rock art. It was very cool and exciting!
Now whilst I said Raman was not chemically specific, one can make it so by coupling Raman spectroscopy with sensing devices based on gold or silver nanoparticles, that have nano-plasmonic properties. This is called surface-enhanced Raman spectroscopy (SERS). When SERS is done properly it can be used in a multiplexed fashion to sense many molecules simultaneously and can generate highly quantitative results. This is not metabolomics per se, as it is now a targeted biochemical assay. You can hopefully see within the metabolomics pipeline that Raman could be a game changer where the analysis of several small molecule biomarkers in population sizes of 10,000-100,000 patients is needed at point of care.
The other area where Raman spectroscopy is exciting is the ability to couple this with microscopy and to image single cells with a pixel resolution of around 500 nm. Raman is not affected by water (unlike infrared spectroscopy) and this method is also non-destructive – two great assets for the analysis of delicate biological material.
I hope that the metabolomics community hears a lot more about Raman spectroscopy in the future.
5. You are part of the UK Consortium for MetAbolic Phenotyping (MAP UK) project. What are the advantages and disadvantages for using metabolomics in large-population cohorts?
I am indeed a small cog within MAP/UK (https://mapuk.org) where our aim is to create a unified platform that will make metabolic phenotyping robust, routine, and reproducible across the UK. I think the advantages are clear. If this MRC funded project is successful then this will be fantastic for the metabolomics community as scientists will be able to work together on common projects, share robust and validated protocols for untargeted metabolomics and targeted assays, and of course more readily share data. This is certainly needed for large-scale multi-centre studies. I actually don’t see any disadvantages in this. I do see the challenges. The most obvious one is to get everyone to agree on specific analytical approaches. For example, many of us use LC-MS for metabolomics, but how many of us use exactly the same stationary phase, and how many of us employ exactly the same gradient elution in the mobile phase. We need adoption of common protocols, or neat computational methods to account for retention time changes between different laboratories, so that metabolite identification is more robust which will lead to more accurate metabolite quantification.
6. You helped established the Centre for Metabolomics Research at the University of Liverpool. How has this shaped your career and what is your advice for early career researchers starting in metabolomics?
It’s being an important part of my academic and research journey. I started off as a bacteriologist in a microbiology department in Bristol and during that time was exposed to analytical methods and data processing. When I moved to Aberystwyth, I got immersed in a group that was really doing exciting high-level computational biology – the first paper we published was on the application of neural networks to address olive oil adulteration. After that I moved to the chemistry department in UMIST in 2003 and I got more heavily involved in analytics and at that time metabolomics start to emerge as an exciting new science. UMIST and the Victoria University of Manchester merged shortly after I arrived in 2004. A couple of years later I moved into the Manchester Institute of Biotechnology where the focus was on interdisciplinary science and brought together scientists from very different disciplines. Although I enjoyed my time in a chemistry department it has been great to have moved in 2018 back into a biological department in the University of Liverpool where the focus is on biochemistry, systems, as well as molecular integrative biology. We have exciting opportunities within the Centre for Metabolomics Research (CMR) and this really drives me. There is great potential to fuse mass spectrometry-based metabolomics with a variety of Raman and infrared spectroscopies that we are developing. Watch this space .
So, what advice would I give ECRs starting in metabolomics. That’s a great question and I think there are many different things I’d suggest anyone to consider. I think the first is to find a research area, perhaps a little bit niche, that one can be passionate about and committed to. Also, to consider that it takes years to establish something that you can feel proud of so patience and perseverance are essential. Developing a thick skin can also help as with every scientific area there will be failures that can be learnt from. Starting anything new will be challenging and so finding a good mentor is critical – I’ve been very lucky in my career to be supported by great mentors. Not only in terms shaping me as a scientist who can be independent, but also grounding me when I get a little too carried away. I think having the right mentor is the key thing. As well as being a great research scientist they should be inspirational and have good leadership qualities. At the same time being a good citizen who cares more about you, the research question, than promoting themselves.