Antony N. Davies,a Mohan Cashyapb and John Hollertonc
aSERC, Sustainable Environment Research Centre, Faculty of Computing, Engineering and Science, University of South Wales, UK
bMASS Informatics, Harpenden, UK
cHollerton Scientific Software Consultancy, St Albans, UK
This is the second part of an interview with John Hollerton, the first part in is the last issue! John recently retired from a long career at GSK and we took this opportunity to have a chat with him, as Mohan Cashyap and our beloved editor Ian Michael both have had the opportunity to work with John on projects and on the LISMS conference (Linking and Interpreting Spectra through Molecular Structures). Here, we move the discussion on to technologies and innovation, the great, the over-hyped and the effectively lost to the modern analytical laboratory.
TD: So, looking at the time span of your career to date, we have seen enormous innovations in spectroscopy, completely different concepts of how to measure spectroscopic data, huge advances in precision and reproducibility of spectroscopic measurements. For you, what was the biggest innovation you have seen which advanced your science?
JH: 2D-nuclear magnetic resonance (NMR) as it has made a massive difference to the whole of NMR. It is hard to overestimate how much the 2D-NMR has changed the NMR technique. Which refers back to my answer in Part 1 of the interview about Ernst & Co. He was a lovely person by the way, I don’t know if you ever got the chance to meet him, but he was the nicest, most self-effacing person that I think I have ever met. Sadly, he died a couple of years ago, but I had the opportunity to meet him and even had breakfast with him once, which was nice! He did a huge amount and he played it down: “Yeah… I did 2D-NMR...”. That is probably the biggest innovation from my perspective. Of course, alongside that has to be powerful computing which enables it. Back in the day when he first did it, it was more an intellectual exercise as there were no computers that were really powerful enough to do 2D-NMR on a regular basis, so powerful computers were a key part to moving things forward.
TD: Of course, there is another side to that coin, which I have experienced with various organisations, and that is what was the biggest hyped innovation which failed to deliver as promised?
JH: In the liquid NMR world, I would say the hyperpolarisation of liquids. People are still working on it, but when it first came out it was considered to be the Holy Grail, because NMR is such an insensitive technique. Hyperpolarisation could give orders of magnitude increases in signal-to-noise, so it was a very laudable aim. However, whilst it has had limited application in liquids, it has turned out to be valuable for solid NMR, so the technique itself is valuable, it’s just that it isn’t great for liquids because of the stress you have to put your sample through in order to do the hyperpolarisation. You have to polarise it at very low temperatures and then suddenly convert it from a solid to a liquid with a rapid temperature cycle. Another issue can be transferring polarisation as it may not enhance uniformly across the molecule, so that integrals become meaningless. With all the negatives it’s rarely appropriate in these cases, whereas it was pushed as the way that NMR would now be competing with Mass Spectrometry for sensitivity. Sadly, you don’t always hit the thing that you’ve been going for—but for solids it’s been great.
TD: and on the computing side?
JH: On the computing side? Well, there have just been so many of them! Almost everything goes through the hype curve, doesn’t it? Everyone jumping on the bandwagon and then finally calming down and becoming more realistic. Cloud computing was originally very hyped up and then it declined as everyone became disenchanted with it, but now cloud computing is a big thing. So, sometimes hype is just things that are before their time. People tend to become too keen and try to ignore the difficulties. In computing there has been a lot of hype around Artificial Intelligence, but AI is now looking promising in many areas. So maybe the key in all these computing things is to use George Box’s quote: “All models are wrong, but some are useful”, so we need to remember that AI may not be reality, but it may be useful.
TD: Looking at instrument development, what were the go-to techniques from the start of your career that are now seldom used?
JH: Well certainly ultraviolet (UV) spectroscopy, as discussed in the first part of the interview. Infrared (IR) spectroscopy is also used far less, in fact, I can’t remember the last time I saw a chemist run an IR spectrum on anything! Electron Ionisation MS (EI-MS) is much reduced, much work is using Electrospray these days because people want a molecular ion and they don’t want fragmentation. Again, it was all originally EI or Chemical Ionisation and now it is probably >90 % Electrospray in terms of volume (mass spec as an HPLC detector).
TD: What innovation would you have liked to have seen before your retirement which hasn’t yet reached fruition?
JH: What I would really like to see is a true set of data standards for analytical data. As you well know with your involvement with JCAMP, it was a very, very powerful and useful thing, and still is—people are still using it today. But I think we have been extremely poor at pushing data standards despite numerous attempts like AniML and Allotrope. It always a chicken and egg situation, if people have it, they will use it. Allotrope have probably got the funding model right as they have a subscription model, so companies pay to belong to Allotrope and that money goes to fund work that is being done by different groups. I was on the Allotrope board for several years and I saw it heading in the right direction, but the trouble is everything takes such a very long time. Everyone has their day job, so it is quite hard to dedicate enough time to it. We also tend to get too fancy with it. One of the nice things about JCAMP was that it was very pragmatic. Often, we are not pragmatic enough and that can derail things in the end. That was probably one of the issues with AniML, as it seemed to get bogged down in the minutiae; maybe taking a step back would have moved things along a lot quicker. Again, it’s just a model of the data, it’s not the truth! So, you must accept that it is not the truth as that would make a huge difference. So, we have a £10 M NMR spectrometer and what we store at the end of the day is a PDF for long-term storage! Is that really the best we can do things in the 21st Century? We should be able to store data in a way that we know we can do something with it in 60 years’ time. So yes, true data standards, I’ve spent quite a lot of time on it, and I hope that it will come… but I’m not confident.
TD: Looking back, did you see any fundamental differences between the various incarnations of GSK?
JH: That’s a really good question! So, yes absolutely. When I joined Glaxo in Ware it was still known as Allen & Hanbury’s and felt more like a family business. Over time, with the various mergers, it became more formal, more impersonal. One indicator of this was the change of title of the “Personnel” to “Human Resources” (HR) department. It says a lot to me about how companies saw their staff. “Human resources” is a very impersonal, “balance sheet”, type of thing, whereas “Personnel” is talking about people. We have lost something important as a result. Some small companies still have it, but big companies generally don’t and that is one of the big disadvantages of working for a big company as you might feel to be a plus or minus on the balance sheet.
TD: That is a very interesting observation about the different titles for the department handling people.
JH: Yes, I don’t know where the term Human Resources came from—possibly out of the US, but it is such a hateful term, as if you’re just running a farm, you talk about head counts… it’s so impersonal.
TD: Was there any difference in access to capital funds?
JH: Well, no not really, access to capital is always tight. Except maybe in my early days at Glaxo when it was growing rapidly and if you wanted something you could probably get it. You just needed to convince somebody of the need. But, over time, there has been pressure to lower costs, because it is an expensive job getting a drug to market. This is also true with operational costs. It is a cost-restrained business. This is what has driven a lot of what I have done. If you know you have limited equipment and limited numbers of people, what you have to try and do is get the most out of the equipment and minimise the time people spend doing things that a computer can do for them. One example is purification factories. Chemists spent a lot of time purifying what they had made. Why not simply automate that process so that they can go back to making more stuff and let someone else worry about purification and registering, as it doesn’t take a chemist to do all that stuff?
TD: Does working in a tightly regulated industry strangle innovation or ensure consistent quality? Although having asked that, I seem to remember that years ago you were happy that you were working in Discovery where there was a much lighter touch if I remember rightly?
JH: Well, I spent time doing projects in Development, so I always had a reasonable understanding of what the compliance requirements were in the Development environment. But coming back to your original question about whether it strangles innovation or ensures consistent quality—the answer is it does both. Its why Discovery is less regulated than Development, because that is where you really want a lot of innovation. It does mean that people are very wary of moving things on, they are very risk averse and often will not take something on unless they can pretty much guarantee it will be a success. So, in that sense it does strangle innovation, and I see it coming more and more into Discovery, which I see with some horror as it will strangle innovation where it really has to happen. I think where regulation is appropriate, it is essential and I think anything that is going to go into humans absolutely has to be regulated, as you can’t afford for there to be mistakes. Ironically if you spend a bit of money in the Development environments to make sure that data flows correctly, again going back to the data standards thing, you get a lot of this stuff for free; so, you need to be persuading people that are doing this fundamental data handling stuff that it will actually improve your compliance.When the regulator comes and says “tell me such and such…” you just press a button and you have your answer rather than “I’m going to go through all my records and going to get this out of this database and this out of this database and put it all together..” you can actually get a lot for free. But, it does require a lot of initial investment, so the question is do I spend the money getting the drugs out of the door or do I spend money so I can get drugs out of the door better in the future? It’s the old story... I’m too busy mopping the floor to find time to turn off the taps!
TD: Now, although we have the habit of concentrating on technologies and software developments, none are effective without a good working environment, which means people. I think, as I have got older and further in different organisations, I concentrated far more on the people rather than the technology. Good people can do wonders with not the best technology, whereas average to poor people can be given the best equipment but still fail to deliver.
JH: Absolutely, I think it is a good point and I see the same as you do. I don’t think companies spend enough time to get people to understand each other and understand their differences. There is a lot of talk about diversity, but it is quite often the wrong diversity—race, religion, gender identity—all aspects of which are important, but there are more fundamental issues around what makes people tick? I’m sure you’ve done the old Myers–Briggs, that was great because it doesn’t matter whether you believe in their quadrants or not, you go through the process and you come out with what my preferences are. And when I was doing it, I was also thinking about people who work for me or work with me. How would they answer that question? And by the time you get to the end of the process you realise why you really do not get on with this person or that: it’s because they are in this quadrant or that and we have very different outlooks on life.
TD: I’m astonished you picked that answer. I have worked in an international management team from quite different backgrounds and experiences that was often at loggerheads with each other. Really difficult to get agreement on anything. Until we undertook the Myers–Briggs training together. I think everyone was quite sceptical at the start. I think we split the workshops over a couple of days and came out of the process completely understanding why Person X found Person Y completely frustrating! So, instead of “why can’t he understand”, or “Why can’t they see this…”. Now I am against putting people in pigeon holes, but it was a great tool for putting people into Myers–Briggs pigeon holes to understand how they functioned and we all got along much better after it, as we now knew “why they couldn’t understand it”.
JH: Yes, in my example I was in charge of a registration team. Their job is to be absolutely focused on the details, so it attracts people who like to focus on the details. But they were turned-off by talk of the bigger picture, which I found very frustrating. So, after I had done the Myers–Briggs I asked them think about the questions as if they were me, and they got me down almost exactly as I had. At the end of it, the head registrar came up to me and said that he now understood that when I said “that’s just the details” it didn’t mean I thought that the details were not important, but just that you were looking at the big picture. Once we understood that, we could see that we are both trying to end up at the same place but using our different skills. I have to say that it was never a bad relationship, but in terms of driving things forward it made a huge difference. People can be sceptical about Myers–Briggs but the key thing is that you understand your differences and you do something about it to use your strengths to get the best result. An ideas person is just that, but you need someone to actually make them work. Someone has to look at the details so you need both sets of people.
TD: If your younger self, just starting out on your career, was attending your retirement party in January, what words of wisdom or warnings would you like to give your younger self?
JH: I would say “Chill and don’t take things personally” … because I used to get incredibly frustrated when people could not see what to me was obvious. I think my behaviour was not great when I started. I think I was arrogant, and I could be short tempered. And I’ve seen your follow-up question about would my younger self have followed that advice—probably not… the arrogance of youth! Persuading people is not about battering them with facts. Sometimes it takes time. My greatest thrill is if I have been banging on about something for a long time and not getting anywhere, to go to a meeting and someone uses my own words back at me as if they’ve just thought of it… Great, we’ve got there! One of the things I’ve learned is if you keep on going head-on against the barriers people put up, you will not achieve what you want but you just have to take a step back and approach things in a slightly different way. I don’t think I would have listened as my younger self; I was far too sure of myself.
TD: From someone with one foot in the University world, I would love your views on whether we are teaching the correct skills to the people you would be recruiting?
JH: In terms of people coming in with degrees or PhDs, I’m not sure that people learn much from spending longer in academia. I think there comes a time when you got to come out and hit the real world. In my opinion the earlier you can hit the real world the better. While I was working at GSK we recruited a lot of apprentices. I think the apprentice approach in the UK is just superb in the right environment. People are brought in and they can do their degree part-time, somewhat going back to the days when you could do a part-time degree. The great thing is that while they are doing the theoretical side of things, they are actually carrying out real practical work and see how the two relate. Almost without fail, these apprentices are the best people that we recruit. By the time they finish their degrees, they are fully functional, very competent members of staff. Back in the day, Glaxo took on a lot of members of staff to do part-time degrees, HNCs or whatever. One day a week to do your academic work takes a lot longer of course, but especially these days you don’t have a massive debt around your neck and businesses end up with the people who have the skills that they want. We should be trying to help Government encourage more of this as they will benefit from all these people getting degrees, whilst being productive for the economy.
TD: So, do you see this academic pairing as important for the skills base going forward?
JH: It tends to be only something we do on things like the apprentice. The standard academic institutions are generally not influenced by industry or if they are, it is very indirectly and particularly in the sciences there should be far more two-way communication about what the academic process is, what things are taught and what things are going to be useful going forward. I don’t think that happens enough currently and it probably needs to happen more. There have been such instances in the past, GSK had a laboratory at Cambridge University in the chemistry department and that really helped cross-fertilise ideas. Unfortunately, that is no longer there, but I know other companies are doing this. Maybe there can be a formal way to engage without actually having to set up laboratories with regular communications between leading employers and academics just to see how relevant the education is. Of course, it doesn’t all have to focused on business as there will always have to be purely academic topics in the curriculum.
TD: I must admit I have been horrified by how many academic institutions teaching chemistry don’t actually let their students touch the analytical equipment. It would seem that the higher-ranking the institution, the more the funding has been funnelled into the research instrumentation, often purchased from research grants and the undergraduates are not even allowed in the room with them. Samples are handed over and spectra appear as if by magic. For me this is a massive mistake. You need to have instruments that people can make mistakes on.
JH: Yes, my example from my experience on the EM360 in the first part of the interview probably formed my thoughts on what my career would be going forward, what stuff I enjoyed doing. But if you don’t get the chance to do this hands-on you will never know until you get into industry and then you might decide it isn’t actually what you want. I think this is important for the sciences, maybe not for the arts, but that’s slightly different and I’m not sure how you would do that, but there are probably equivalent things in the arts. Having that link to the world outside academia is really good.
TD: What are your plans for the future? You’ve already mentioned you are doing some consulting?
JH: I am doing some consulting and I will continue as long as I find it interesting and am adding value. I do like to have time to do other things, which is fine. I would liked to have continued with Allotrope, but the way Allotrope is set up makes that almost impossible for an individual to contribute. This is something I may bring up with them at some point. I still see Allotrope as being the best chance we have at analytical data standards, but it has taken a long time. It always takes a long time for these things but it has taken a very long time so I would like to still have some influence on it.
TD: My worry about this and several other similar initiatives has been the fact that they wouldn’t put anything out into the public domain. That made it almost impossible to help or critique or whatever. Often all you could get out of them was “we are working on it” to questions about how are you dealing with this? So, how’s the documentation going… how far have you got... there is just no feedback. There was a lot of good, expensive marketing at a number of LIMS-type conferences, but in the end you couldn’t get involved in any discussions of substance or try to align the work you yourself are doing.
JH: I think Allotrope has changed as originally it was all about the funding model and I think the assumption was if you are going to pay money for it you need to get something out of it, so it was a closed shop. The reality is I think most companies would have joined even if they hadn’t got exclusive rights on what was being produced. So, things like the ontologies are public domain, which I think is good and a step in the right direction. The simple data model they have got, which is a simple JSON representation of the data, will become public domain de facto as you can reverse engineer it. Where it was originally with the binary representations which required the libraries to do anything with it—well I was never a great fan. I understood why they did it, but it was too complex and you were required to have a very in-depth knowledge of how the data structure was set up and what the bits mean.
TD: This reminds me of discussion in the ASTM AniML committee where the programmers said they would only store 8-bit numbers—we pointed out that many instruments were storing 16-bit natively, so they came back and said we will only store 16-bit numbers... so, we said what if the NMR instruments are storing 64-bit data? You simply cannot restrict the formats in these ways. You must be flexible enough to be able encode whatever the measuring system is providing.
JH: Yes, it’s one of the areas where I have been in numerous discussions on the best way forward and the bottom line is it depends on what sort of data you have. The idea about Allotrope is that it is meant to be a data standard to cover all scientific data. Now, obviously that is a very big thing and its not going to happen straight away, and you need to concentrate on certain areas, but it should be capable of being extended to other things in the future. The first thing is to make sure you call the same things by the same names and that is the ontology. Next a lot of the data can be represented in ASCII, so you could put that in JSON, but very large data sets like TOF MS are too huge, so you have a companion file which is a binary representation of the original data which hooks into your JSON file. So, if you are only interested in the metadata, you can simply get that from the JSON ASCII file; but if you want to read the data you will need the binary file and some tools to do that. So, I pushed very hard while I was there for anyone with elementary programming skills to be able to say I’ve got an Allotrope JSON file and I know how to parse it; in fact I already have the framework to parse it. It should be democratised to that extent and that was the great thing about JCAMP—anyone could work out how to parse a JCAMP file. I know when JCAMP came out and I was working on vibrational spectroscopy I could parse the files and write my own peak picking algorithms as the ones on the instrument were a bit rubbish! So, I could properly identify shoulders and represent it all in a plot and all that stuff I could do because of JCAMP; otherwise it would have meant dealing with vendors’ binary formats that they would not publish. So that was probably one of my experiences that made me so keen to get good analytical data standards and one of my biggest regrets that I didn’t manage to do that. Maybe I left it in a slightly better place than when I arrived.
TD: So, on that positive note is there anything else you would like to discuss before we put a wrap on this interview?
JH: Well, I think it’s been pretty extensive. It’s brought back many memories of things I did when I was just starting out. I hope it will be interesting for people to read. Going back to the life lessons, I heard Digby Jones, once head of the Confederation of British Industry, say he used the phrase SUMO—Shut Up and Move On. He said that there are times when you’ve tried your best and you know that it’s not getting anywhere. Shut up. Move on. There are other things to do. He gave an interview about his history and that is one of the things that really stuck with me. It doesn’t mean giving up on something it just means stop being so vocal about it, move on to something else because it will come around again. If it is important, it will come around again!
TD: So, it is definitely time to SUMO!” Thanks for your patience and I hope it wasn’t too intrusive.
