1 00:00:00,870 --> 00:00:04,130 Can you just give me your name and say what? Your title here? Yes. 2 00:00:04,560 --> 00:00:09,390 My name is Brian Marston, an associate professor for Data Management and Research Informatics, 3 00:00:09,870 --> 00:00:12,960 and I'm also associate head of Division for Digital Information. 4 00:00:13,380 --> 00:00:19,150 Great. And I don't think you mentioned what Unity were attached to, because I'm attached to multiple units. 5 00:00:19,220 --> 00:00:26,790 Yes. So I'm attached to the Kennedy Institute for Haematology and Dorms and the Centre 6 00:00:26,790 --> 00:00:30,300 for Medicine's Discovery and DM and the Welcome Centre for Human Genetics in India. 7 00:00:30,690 --> 00:00:36,270 Right. Great. Thanks very much. So can you just start by telling me a little bit about yourself, starting from what? 8 00:00:36,300 --> 00:00:39,500 How you first got interested in science? Yes. 9 00:00:39,840 --> 00:00:46,350 Pretty like the main staging points in your career up to until the end of 2019, 2018. 10 00:00:46,350 --> 00:00:51,270 Yes, that seems like a long time ago. So, yes, science has always been my thing. 11 00:00:51,270 --> 00:00:56,490 My father was a scientist. He studied at Leeds, the structure of penicillins. 12 00:00:56,970 --> 00:01:03,690 Back in the fifties and sixties. There was an expectation that I would be a scientist, but nevertheless it came out that way. 13 00:01:04,260 --> 00:01:09,570 So I did my undergraduate degree in natural sciences at Cambridge, focusing primarily on biochemistry, 14 00:01:10,140 --> 00:01:19,379 then came here to Oxford to do a dphil in while biochemistry, looking at extracellular matrix proteins and the dynamics of those within. 15 00:01:19,380 --> 00:01:23,070 Campbell Then I moved to San Diego. 16 00:01:23,820 --> 00:01:29,070 I worked at the Scripps Research Institute for a number of years, and then I kind of focussed more on the computational side of things. 17 00:01:29,070 --> 00:01:30,900 So that's where you first got interested in? 18 00:01:30,930 --> 00:01:38,100 Yeah, they actually happened in the undergraduate end of things because we soon found out I was very dangerous in the lab. 19 00:01:38,430 --> 00:01:46,440 And so there was a what is called a Silicon Graphics machine, which is a very fancy high end PC effectively corner with no one's using. 20 00:01:46,440 --> 00:01:49,680 And I managed to get it to work. And the rest, as they say, is history. 21 00:01:50,490 --> 00:01:52,950 So as a consequence, I have, you know, the, 22 00:01:53,280 --> 00:01:59,130 the wet lab experience as well as the computational experience and you can speak both those languages is a little bit unusual. 23 00:02:00,030 --> 00:02:08,310 So yes, the scripts where I help them build some of the first high performance compute clusters over there worked in the lab of Ruben Avigan, 24 00:02:09,480 --> 00:02:18,660 who was interested in computational chemistry. So I built a number of algorithms and then I moved to industry for a number of years by a focus where 25 00:02:18,660 --> 00:02:24,680 I led the computational chemistry team on the design of ion channel and kinase inhibitors and well, 26 00:02:26,040 --> 00:02:30,030 briefly for for the ignorant, what is computation? It's a great question. 27 00:02:30,170 --> 00:02:35,040 It's a computational chemistry. It depends on who you are and how you define it inevitably. 28 00:02:36,150 --> 00:02:45,690 So you can imagine that there is a computational chemistry in terms of the fundamental understanding of how small molecules, 29 00:02:45,690 --> 00:02:50,640 particularly organic molecules, how what their shapes may be, what their properties might be. 30 00:02:51,600 --> 00:02:53,759 And that's not so much on the drug discovery pathway. 31 00:02:53,760 --> 00:03:00,360 That's more like pure theoretical chemistry and use computation to be able to calculate these values and simulate them. 32 00:03:00,780 --> 00:03:02,280 That's not what I do. 33 00:03:02,970 --> 00:03:14,940 What I do is focusing more on using computation to better understand ways to develop small molecules or biologics in a drug discovery pipeline. 34 00:03:15,990 --> 00:03:22,650 And so this involves not just the small, organic, small molecule itself and the drug, 35 00:03:22,950 --> 00:03:28,800 but also the protein that that small molecule may actually want to bind to and how it may inhibited and so on and so forth. 36 00:03:29,610 --> 00:03:32,250 And so that's much broader, tends to be a bit more practical. 37 00:03:32,610 --> 00:03:40,740 So it really is a case of in an industry you might do something called a high throughput screen where you screen many millions 38 00:03:41,070 --> 00:03:46,830 of these small molecules against the protein you're interested in you think is associated disease and you get some hits. 39 00:03:47,160 --> 00:03:50,430 And the question then is can one then modify those hits, 40 00:03:50,430 --> 00:03:54,149 those small molecules so that they're more potent and selective so in other ways 41 00:03:54,150 --> 00:03:57,560 that they stick better to your protein and they only stick to your proteins, 42 00:03:57,570 --> 00:04:01,960 you don't have side effects. And that's partly due to your screening in the lab or computationally both. 43 00:04:02,340 --> 00:04:09,600 So that's at the screening. In the screening, the lab is done at the Centre for Medicines Discovery or with our partners in industry, 44 00:04:09,600 --> 00:04:13,620 and then the results from that then come to us for then to age computation, 45 00:04:13,620 --> 00:04:17,040 interpret it and to make suggestions because they're not always right, 46 00:04:17,040 --> 00:04:23,070 but suggestions to the chemists about how they may modify those small molecules to make them better. 47 00:04:23,760 --> 00:04:29,690 And that's the beginnings of the drug discovery process. So I guess I'm what's the it's just drill down a little bit. 48 00:04:29,760 --> 00:04:33,450 What what is the data that you're putting into your algorithms? 49 00:04:33,450 --> 00:04:37,540 Yeah. So as I said, there's the high throughput screening. 50 00:04:37,540 --> 00:04:42,959 Yes. Whereby you know what your small molecule is and you know whether it had some 51 00:04:42,960 --> 00:04:48,480 impact upon the assay that you're using within the actual throughput screen. 52 00:04:49,020 --> 00:04:54,390 And then you can start looking at the shapes and the structures and the properties of these small molecules. 53 00:04:54,750 --> 00:04:59,400 You might start seeing patterns, You might see see patterns about particular. 54 00:04:59,820 --> 00:05:05,530 Types of substructures to be columns of particular atoms or collections of atoms on that molecule. 55 00:05:05,850 --> 00:05:12,590 That might mean that they are more sticky or they are more polar, or they are something, or they behave in a certain way electronically. 56 00:05:12,600 --> 00:05:20,460 And at that point you can try and correlate that with an understanding of the protein that they bind to. 57 00:05:20,790 --> 00:05:24,500 So the analogy I tend to use is that of Lego bricks. 58 00:05:24,510 --> 00:05:27,780 I mean, when we solve the structure of a protein, 59 00:05:27,780 --> 00:05:35,940 we are effectively writing the manual for a LEGO model and we're kind of working out where all the bricks go, what colours, what size, what shapes. 60 00:05:36,840 --> 00:05:41,190 And then when we have a small molecule, we've got another piece of Lego sets, a piece of Lego on. 61 00:05:41,190 --> 00:05:50,340 What we have to do is design those sets, a piece of Lego to fit as much as possible into the cavity that exists in the main Lego protein model. 62 00:05:51,060 --> 00:05:59,070 So that's that's what this is all about. It sounds very visual and even very 3D, the way you very 3D use it, very thin. 63 00:05:59,670 --> 00:06:08,969 I think what I'm trying to get at as a non information scientist is how you do that computationally how 64 00:06:08,970 --> 00:06:15,270 you describe the entities are they described in terms of a three dimensional shape or in the forces? 65 00:06:15,480 --> 00:06:18,210 Yes, All all of all of those those things, yes. 66 00:06:18,330 --> 00:06:25,290 So yeah, the 3D structures, sentiments, discovery, we we solve the structures of human proteins often for the first time. 67 00:06:25,290 --> 00:06:31,710 No one's ever seen them before. And from that, we can then start to understand how those proteins actually work. 68 00:06:31,830 --> 00:06:39,840 So the mechanisms and mechanics of them, we can also correlate that with no mutations. 69 00:06:39,840 --> 00:06:47,010 So we might know that in a particular position in the protein sequence, there is a change in what is called the residue. 70 00:06:47,430 --> 00:06:51,570 And that residue means that there's a change in one of the Lego blocks effectively. 71 00:06:51,900 --> 00:06:57,780 So at that point you could then look in your protein structure in 3D and you can rotate it around on a computer screen and you can see how either 72 00:06:57,780 --> 00:07:03,659 the change of the colour or the change of the shape of that particular Lego brick is going to have an impact on what that protein actually does, 73 00:07:03,660 --> 00:07:07,390 what it switched on cause it switched off less. You know, all these sorts of things. 74 00:07:07,410 --> 00:07:12,540 So yes, that's, that's how that works. Yeah. So I completely interrupted your life story. 75 00:07:12,540 --> 00:07:16,310 So we got as far as scripts and you said that was. Yeah. Into the computational. 76 00:07:16,320 --> 00:07:20,250 Yeah that's right. So then then moved to by focus that. 77 00:07:20,670 --> 00:07:28,000 Oh yes, Yes that's right. Yeah. And then work there for two and a half years and then got very frustrated with the, 78 00:07:28,170 --> 00:07:32,610 the, the lack of blue sky thinking and the risk averse nature of industry. 79 00:07:33,300 --> 00:07:41,850 And I got called up in 2004 by a person called Michael Sundstrom who helped co-found the SGC here. 80 00:07:42,210 --> 00:07:46,320 And he said, We need someone to come and do something called research Informatics. 81 00:07:46,990 --> 00:07:48,959 I thought, Well, we'll have a discussion about it. 82 00:07:48,960 --> 00:07:57,180 So I came over and he he sat me down and, you know, sat in front of the PowerPoint slides, and he ran through the PowerPoints. 83 00:07:57,180 --> 00:08:00,569 And it became very clear that what they were trying to do is highly ambitious. 84 00:08:00,570 --> 00:08:05,610 They were trying to solve the structure of human proteins, had never been done before at scale. 85 00:08:05,610 --> 00:08:09,540 So hundreds of years. And this was something that no one thought was possible. 86 00:08:09,750 --> 00:08:13,920 This is just after, you know, the human genome sequence. So everybody wanted to get this done. 87 00:08:14,310 --> 00:08:18,150 So you said SGC, this Structure Genomics Consortium. 88 00:08:18,620 --> 00:08:20,429 It was, well, let me explain. 89 00:08:20,430 --> 00:08:30,750 So as I said, the human genome was solved to an extent, and we kind of then knew what proteins could be made in a human cell. 90 00:08:31,530 --> 00:08:35,099 And so the question then was for the pharmaceutical companies, well, 91 00:08:35,100 --> 00:08:38,700 how can we leverage this information to make smart decisions about drug discovery? 92 00:08:39,390 --> 00:08:46,440 And that the existential risk was that there were a number of new Start-Up companies dotted around the world, 93 00:08:46,440 --> 00:08:49,830 but particularly on the west coast of the US, which said, 94 00:08:50,100 --> 00:08:54,089 well, we've got access to human genome, we're going to solve the structure of the human protein, 95 00:08:54,090 --> 00:09:00,270 I'm going to license them to you for large amounts of money. And the pharma companies said this is not a good thing. 96 00:09:00,750 --> 00:09:09,000 What we need is a pre competitive effort that is going to solve these structures, put them out of the public domain without any burden of use. 97 00:09:09,000 --> 00:09:13,650 You know, you don't pay for it, it's just freely available. And that led you to that problem and everything is good. 98 00:09:14,880 --> 00:09:20,550 Now it turns out that solving these human protein structures is a heck of a lot more harder than anyone thought it was going to be. 99 00:09:20,940 --> 00:09:26,459 So back in 2000, this originally sort of kicked us around about 2000, 2001. 100 00:09:26,460 --> 00:09:33,270 At that point, there were about seven or eight farmers involved. But by 2003 became clear this wasn't such an existential risk. 101 00:09:33,780 --> 00:09:40,260 And then there was only GSK involved and um, and the welcome or welcome trust as it was then. 102 00:09:40,740 --> 00:09:46,920 And the Wellcome Trust wanted to leverage its investment into the, you know, the solving of the human genome. 103 00:09:47,250 --> 00:09:49,229 And that seemed like the next big step. 104 00:09:49,230 --> 00:09:55,710 So they said, well, we need to set up a pre competitive structured genomics consortium, which is what they did here in Oxford and also in Toronto. 105 00:09:56,130 --> 00:09:59,190 So yes, so I then came in and helped to. 106 00:09:59,280 --> 00:10:03,120 Build mostly the data management and some computational chemistry around that. 107 00:10:03,420 --> 00:10:09,180 And the idea was that the SGC would make all these humans and we did. 108 00:10:09,540 --> 00:10:13,560 Yes. Yup. And make them. It was an open access project. 109 00:10:13,590 --> 00:10:21,360 Totally open access, Yes, exactly. And, you know, I, I think at the last count, we've sold over 2000 of them, and we are the world leaders on this. 110 00:10:21,840 --> 00:10:25,020 Um, so. Yes, very exciting. Mm hmm. Mm hmm. 111 00:10:25,350 --> 00:10:34,630 And the the technology to solve those structures was, again, partly computational and partly using things like the diamond light source to. 112 00:10:34,800 --> 00:10:41,010 Yes, well, diamond didn't exist in those days, so we had to use a place called SRF in France. 113 00:10:41,010 --> 00:10:44,250 But there is some computational aspect to that. 114 00:10:44,670 --> 00:10:51,780 That was not what I was responsible for. As part of the solving of protein structures, you can use a method called crystallography, 115 00:10:52,410 --> 00:10:56,580 which is where you have your protein very pure and very small amounts of it, 116 00:10:57,060 --> 00:11:06,270 and you put it into something called a crystallisation experiment whereby you try and coax the protein to turn into a crystal, a very uniform crystal. 117 00:11:06,540 --> 00:11:15,000 These crystals is often smaller than the naked eye can see, but yet you can shoot very strong X-ray beams through them, 118 00:11:15,180 --> 00:11:18,960 which from a synchrotron like SRF of diamond, 119 00:11:19,440 --> 00:11:25,260 and from that you get diffraction of the X-ray beams according to how the atoms are placed within your protein. 120 00:11:25,650 --> 00:11:28,110 So using some very fancy maths and some computation, 121 00:11:28,380 --> 00:11:34,320 you could then infer roughly where all the atoms are and therefore you know where your Lego bricks are and therefore you've got the structure. 122 00:11:35,310 --> 00:11:39,000 So there is some computation involved with that, but not as much as you might imagine. 123 00:11:39,000 --> 00:11:44,010 Actually, the hard step is actually making the protein, making sure it's stable, 124 00:11:44,010 --> 00:11:47,280 making sure you can get the crystals and actually solving a structure. 125 00:11:47,280 --> 00:11:50,790 It actually is not that hard in the grand scheme of things. So what was your role? 126 00:11:50,790 --> 00:11:55,919 Is that your role? Yeah, So so SGC, we wanted to do things a little bit differently. 127 00:11:55,920 --> 00:12:02,010 We knew we were an academic organisation, but yet we have industry milestones and deliverables in the sense that, 128 00:12:02,400 --> 00:12:07,050 you know, our funders expected us to deliver X structures per year, right? 129 00:12:07,260 --> 00:12:15,660 Or actually X structures per month actually. Um, and we knew that we, we couldn't do everything in a standard academic process. 130 00:12:16,140 --> 00:12:18,750 And so a number of us that were hired, I was employed like number three, 131 00:12:18,750 --> 00:12:26,370 I think there were four or five of us who came from industry and knew how to work in a process oriented kind 132 00:12:26,370 --> 00:12:33,750 of approach and in an industrial approach in terms of making sure you meet the deliverables in the mountains. 133 00:12:34,200 --> 00:12:39,180 And as part of that, it became increasingly clear that what we needed to have was an appropriate 134 00:12:39,420 --> 00:12:44,640 data management infrastructure underneath that to capture information about, 135 00:12:45,000 --> 00:12:51,450 um, what proteins were working on, what ways we have tried to make the protein, um, 136 00:12:51,450 --> 00:12:57,179 what ways we've tried to solve the structure of the protein, what work and all that sort of infrastructure around that. 137 00:12:57,180 --> 00:13:03,780 So electronic lab notebooks, databases and so forth. So that's what I built over the first four or five years within the SGC, 138 00:13:04,080 --> 00:13:08,580 and that was used in Oxford, but also the other SGC nodes around the world as well. 139 00:13:08,580 --> 00:13:14,580 So yeah, so I that, that was the point at which I tend to move away from the computational chemistry and more to the 140 00:13:14,970 --> 00:13:19,620 what I call the standard data management expects to work I guess in a library you call meta data. 141 00:13:20,100 --> 00:13:26,610 Well, yes, that's right. So metadata being information about data, right? 142 00:13:26,970 --> 00:13:32,610 Yeah. And making sure you're capturing as much metadata as possible because you can never go backwards. 143 00:13:34,620 --> 00:13:38,820 And so, I mean, what what were your main concerns? 144 00:13:39,570 --> 00:13:48,310 Well, I came to the timing is interesting. Because you've I mean, you moved from the SGC to the. 145 00:13:51,770 --> 00:13:55,080 Centre for Drug. Sorry, I can't keep you. No, let. Let me expand this. 146 00:13:55,110 --> 00:14:01,500 Been a bit. There's a bit more history. So the CDC here not said existed until 2020? 147 00:14:01,520 --> 00:14:09,010 Yes. And in 2020, the university made the decision that we should no longer be part of the CDC. 148 00:14:09,380 --> 00:14:12,740 Various reasons. And so but we still exist. 149 00:14:13,400 --> 00:14:17,300 And so we had to relaunch ourselves. And now we're known as the Centre for Medicine and Discovery. 150 00:14:17,660 --> 00:14:23,090 We're still is still the Oxford node. That's right. But we are not we're not affiliated with SGC anymore, 151 00:14:23,330 --> 00:14:29,030 but we are still doing CDC like things in the sense that the science is the same and the ethos is the same, 152 00:14:29,030 --> 00:14:34,430 and that we are still very strong proponents of open access science and, 153 00:14:34,670 --> 00:14:41,959 you know, making reagents and know how available to other scientists to really catalyse the drug discovery process and the data centre, 154 00:14:41,960 --> 00:14:48,000 because I tend to talk to people about what they did up to the end of 2019 and then January 2020. 155 00:14:48,020 --> 00:14:54,260 COVID is on the horizon and so that changes everything. But that you were saying this change happened during 2020. 156 00:14:54,590 --> 00:14:58,280 Was it already in the wind before the pandemic came along? 157 00:14:59,450 --> 00:15:07,759 It was mooted probably it became clear that it was a possibility around November, December 2019, I think. 158 00:15:07,760 --> 00:15:11,960 And then there were a lot of very difficult conversations that occurred over the following period 159 00:15:12,320 --> 00:15:17,390 and how that was going on and how let's just cover this now and we'll get on to it in a minute. 160 00:15:20,010 --> 00:15:25,790 Who who are your customers? Essentially, because you work as a essentially as a service to others? 161 00:15:25,800 --> 00:15:37,700 Well, what we do and we don't yeah. So we we we are not like the SGC back in what we call phase one, which is years 24 to 27. 162 00:15:37,700 --> 00:15:40,909 And then we had a single mission which was sold the structures, that was it, 163 00:15:40,910 --> 00:15:45,170 There was nothing else is not right papers or bringing grant funding that was. 164 00:15:45,800 --> 00:15:49,160 But of course, you know, that cannot last. 165 00:15:49,340 --> 00:15:57,229 And we were fortunate to be funded, you know, four times by welcome, three times by up to nine pharma companies at any one moment. 166 00:15:57,230 --> 00:16:02,120 You know, that's a heck of a lot of money. But those days have moved on and that's not possible anymore. 167 00:16:02,120 --> 00:16:08,930 And so we now are much more academic in the sense that we are having to bring in the grant funding to do what we want to do. 168 00:16:09,680 --> 00:16:15,650 And there are clear academic projects in the classical sense that are being run. 169 00:16:16,010 --> 00:16:22,460 Having said that, we still retain a lot of the best things, particularly around platforms. 170 00:16:22,850 --> 00:16:31,489 So we have what the university calls small research facilities around protein crystallography, around what we call biotechnology. 171 00:16:31,490 --> 00:16:38,660 So that's the production of the proteins around screening of the proteins with small molecules and around the data management actually. 172 00:16:38,990 --> 00:16:44,420 And those are self-sufficient in the sense that they provide services to the rest 173 00:16:44,420 --> 00:16:48,740 of the CMG as well as of the people in the university and industry as well. 174 00:16:49,130 --> 00:16:58,070 So our customers are ourselves, um, local collaborators, local biotech and semi and industry. 175 00:16:58,400 --> 00:17:06,740 And actually the data that we're producing collectively with CMT, our customers, the world, because you know, the structures, 176 00:17:06,740 --> 00:17:15,020 the know how, the expertise that we put out freely available, we know is highly prized by people because they know it is reproducible. 177 00:17:15,890 --> 00:17:20,570 And you're probably aware that there's a bit of a reproducibility crisis in academic science, particularly discovery science. 178 00:17:21,110 --> 00:17:25,969 Um, one of the reasons why we, we got it to the point where we had nine farmer funding at any one moment and they 179 00:17:25,970 --> 00:17:30,370 kept coming back was because they knew that if they asked us to do something, 180 00:17:30,380 --> 00:17:34,430 solve a structure that they were interested in, they would be able to replicate it first time in their lab. 181 00:17:34,430 --> 00:17:36,680 And if they couldn't, they could pick up a phone and talk to us. 182 00:17:37,730 --> 00:17:46,549 And so that has been transformative to many of our farmer partners because they can never really describe what the actual impact is, 183 00:17:46,550 --> 00:17:51,890 because, you know, they're not gonna tell us how many proprietary programs we've we've enabled, but we know we've enabled very many. 184 00:17:51,890 --> 00:17:57,030 We know we've enabled a number of drugs and critically we know that we have, um, 185 00:17:57,350 --> 00:18:05,450 provided data or capability to generate data that has terminated projects early, which in industries particularly important, saves money. 186 00:18:05,450 --> 00:18:08,620 Yes. Yes. And time and duplication of effort. Yes. Yes. 187 00:18:08,630 --> 00:18:12,110 Yes. Yeah. So let's finally arrive at COVID. 188 00:18:12,110 --> 00:18:20,090 I'm asking everybody this. Can you remember where you were when you first heard that there was something happening in China and and 189 00:18:20,330 --> 00:18:26,830 how soon it was before you realised that it was going to be serious and be it was going to involve you? 190 00:18:27,350 --> 00:18:31,790 Well, so I don't remember where I was when it really started to pick up. 191 00:18:32,300 --> 00:18:36,200 I mean, it became clear early March that something was going on. 192 00:18:37,670 --> 00:18:43,399 Obviously by mid-March it was clear something was going on and we didn't really know what that meant for us. 193 00:18:43,400 --> 00:18:47,210 We were just, you know, there was lots of concerns, understandably. 194 00:18:47,840 --> 00:18:51,140 Um, and one of my bosses, chest bound. 195 00:18:51,530 --> 00:18:54,710 Brian, you need to go and help out in this time other than what people want. 196 00:18:55,550 --> 00:19:00,379 And then it was on a Sunday in the first week of April, April the fourth or something like that. 197 00:19:00,380 --> 00:19:05,700 And Dave Stewart called me up and said, Brian, we need you. 198 00:19:05,720 --> 00:19:11,360 Okay, What do you need? We need a data platform because we're going to do a serology platform. 199 00:19:11,780 --> 00:19:15,070 We need it in two weeks. Okay. 200 00:19:16,220 --> 00:19:23,510 Would it normally take you? Oh, about nine months. And I said, Dave, you do realise it's going to take nine months? 201 00:19:23,720 --> 00:19:27,590 We don't have time for that. You just got to do what you can. This is an international emergency. 202 00:19:27,740 --> 00:19:32,540 That date. I get it. But I'm just warning you, I'm trying to. And of course, it didn't take us two weeks. 203 00:19:32,540 --> 00:19:37,710 It took us six or seven. But nevertheless, that's when it all really kicked off. 204 00:19:37,760 --> 00:19:42,650 And then life became very, very busy, very, very quickly. And what was that platform set up to do? 205 00:19:42,800 --> 00:19:48,050 So I have spoken to Dave. I have spoken to Terry. Let me let me give my perspective. 206 00:19:48,410 --> 00:19:51,170 My perspective. So as you will remember with the discussions for them, 207 00:19:51,170 --> 00:19:59,840 this is a platform to take participants or patients blood samples and to determine how 208 00:19:59,840 --> 00:20:05,000 much what what to what degree they had antibodies against the spike protein in COVID 19. 209 00:20:05,480 --> 00:20:13,190 And yeah, so it became very clear that the government needed a way to do this in a kind of surveillance mode. 210 00:20:13,670 --> 00:20:18,200 And it became very clear that Office national Statistics was going to be involved. And then Sarah Walker got involved. 211 00:20:18,200 --> 00:20:22,250 And, you know, at that point it's like, well, there's half a million in the cohort. 212 00:20:22,400 --> 00:20:28,730 It's going to be about 6000 samples a day. Thank you very much. And it's going to may even be a 24 seven things is what we thought at the beginning. 213 00:20:29,780 --> 00:20:35,659 And then, of course, we we had to commandeer the the robotics in the Target Discovery Institute for Dan and his group. 214 00:20:35,660 --> 00:20:38,750 And there were lots of problems associated with that. No reflection upon him. 215 00:20:38,750 --> 00:20:48,559 It's just really hard to set up. And what we had to do and I'm fortunate my team at the same D is just so flexible is we were not in a position to 216 00:20:48,560 --> 00:20:54,440 be able to define what the requirements were for the data management around that because it was changing daily. 217 00:20:54,980 --> 00:21:00,500 We knew we were going to have to work with the Office for National Statistics, 218 00:21:00,680 --> 00:21:05,900 with the company IQ view that was dealing with the logistics of receiving the samples and getting them to Oxford. 219 00:21:06,230 --> 00:21:09,380 You see that name again, IQ, IQ VII. 220 00:21:10,020 --> 00:21:14,510 So they were the company that worked with the government to do the logistics around, you know, 221 00:21:15,080 --> 00:21:19,190 it was originally going to be Amazon and it didn't work out and Royal Mail and so on and so forth. 222 00:21:19,190 --> 00:21:25,700 So it was very complicated in those days. And so we had because we would we had to know what the samples were that were coming. 223 00:21:26,330 --> 00:21:30,890 And so that when we reported the samples back to the Office of National Statistics and the UK Government, 224 00:21:31,370 --> 00:21:36,620 we were, you know, using the right identifiers so that they could then merge that data back into what they had. 225 00:21:37,070 --> 00:21:45,320 So it obviously became very complicated in three ways in the sense that we had to be able to take data 226 00:21:45,320 --> 00:21:51,380 in on a daily basis from a third party organisation who were in just as much blind panic as we were, 227 00:21:51,890 --> 00:22:00,379 and we had to provide data to an organisation which were also just in as much of blind panic as we were and lots of discussions around them. 228 00:22:00,380 --> 00:22:06,650 And they do not necessarily speak academic academia and we do not speak government or industry. 229 00:22:07,130 --> 00:22:12,250 And so one of my roles was to try and have these conversations and to develop these conduits for this data. 230 00:22:12,260 --> 00:22:13,880 So that was challenging. 231 00:22:14,420 --> 00:22:20,780 But the most challenging thing was we actually didn't know why I said it was going to be we knew was going to be analyser assay. 232 00:22:20,780 --> 00:22:24,649 We knew that that's fine, but we didn't know how it was going to operate. 233 00:22:24,650 --> 00:22:29,720 We didn't know how the data was going to come off. We didn't know how to make decisions about quality control. 234 00:22:29,960 --> 00:22:35,960 We didn't know how to make decisions about how to analyse the data, and we didn't really know how to report it. 235 00:22:36,560 --> 00:22:40,010 And this we were learning on the fly. And that all sounds great, right? 236 00:22:40,010 --> 00:22:44,960 But what we had to build was a database and a web platform that sits on top of it, 237 00:22:45,380 --> 00:22:49,700 which everyone across the whole team in Oxford would have access to, 238 00:22:49,700 --> 00:22:58,150 to be able to, you know, upload the data and be able to track how things were going, check the quality control and so on and so forth. 239 00:22:58,520 --> 00:23:04,429 And I'm very fortunate that, you know, the couple of people on my team who did this pretty much 24 seven for the first 240 00:23:04,430 --> 00:23:10,190 couple of weeks were really open to the idea of things changing all the time. 241 00:23:10,190 --> 00:23:13,370 So, you know, they'd write a bit of code and it was right for yesterday. 242 00:23:13,760 --> 00:23:17,300 But this morning we've changed things slightly and that code is going to have to change. 243 00:23:17,720 --> 00:23:21,770 And in the normal circumstances, my team will get very frustrated by that because it's a waste of time. 244 00:23:21,770 --> 00:23:26,180 But this is just a function of how things were operating at that stage. 245 00:23:27,230 --> 00:23:32,540 So normally you would have done all that and prep and testing and setting up to even start it. 246 00:23:32,540 --> 00:23:36,140 Yeah. And then we would have to sit down and say, well, you know what you want to do now. 247 00:23:36,350 --> 00:23:43,399 So just tell us precisely, you know, how you want to catch the data, work with data, present the data and we'll build something around. 248 00:23:43,400 --> 00:23:48,500 You know, this was this was a deep collaboration where they were saying, now we're going to do it this way now. 249 00:23:49,200 --> 00:23:54,849 And we're saying, well, we can't. You sing that quick. But what if we do it this way and then they change their approach? 250 00:23:54,850 --> 00:23:58,120 So there was a lot of to and fro, and that worked incredibly well. 251 00:23:58,280 --> 00:24:02,370 Mm hmm. And how long did that? Six. 252 00:24:02,370 --> 00:24:05,620 6 to 7 weeks. Yes. Did you get the initial thing up and running? 253 00:24:05,650 --> 00:24:13,210 Yeah. To be fair, we probably had something working within four weeks, but the the the on the data management. 254 00:24:13,630 --> 00:24:17,400 But on the actual assay platform itself, it's more like 6 to 7 weeks. 255 00:24:17,410 --> 00:24:25,450 So it was more like end of May, early June, before we really had something that we could go to owners and say, It's good, let's go. 256 00:24:25,690 --> 00:24:31,719 Mm hmm. And so was it from that point that the UN started publishing that national data So you'd get that? 257 00:24:31,720 --> 00:24:35,350 Yes, exactly. Week. Yeah. Yeah, exactly. And did the platform have a name? 258 00:24:35,350 --> 00:24:41,710 Just Oh, the data platform is called Eliza LIMS. Right. Because I couldn't think of any better library information management. 259 00:24:42,000 --> 00:24:47,740 No. So yeah, so there's Eliza and then LIMS stands for Laboratory Information Management System, also known as Electrum member. 260 00:24:48,370 --> 00:24:51,870 But it was bespoke. Um, yeah. So. Mm hmm. 261 00:24:52,160 --> 00:24:56,470 Mm hmm. But that was only the first project, was it not? 262 00:24:56,500 --> 00:25:03,310 Oh, yes. Well, yes, and there have been a couple of projects, so let's talk a little stick on that one and just say that, 263 00:25:03,790 --> 00:25:09,110 um, you know, the little platform is now used out in Thailand at Morro. 264 00:25:09,910 --> 00:25:13,180 They've been using that for their Oxford research. 265 00:25:13,180 --> 00:25:21,999 Yes, it is. So that's that's been a real success for us, is being able to take what we built with lots of sticky tape and elastic bands and to be 266 00:25:22,000 --> 00:25:26,889 able to transplant that somewhere else and get it to adapt to the way that they're working, 267 00:25:26,890 --> 00:25:31,120 which is very different from the way that we were working here. The volume is different, expectations are different. 268 00:25:31,540 --> 00:25:37,600 That's been a huge success. And there's they're also looking at coded and other pathogens as well. 269 00:25:37,900 --> 00:25:44,200 In terms of malaria. Yes, I mean, they've just finished going through the COVID, but there are other pathogens that they're starting to look at now. 270 00:25:44,200 --> 00:25:49,060 So I think that's been a huge success, too. So there's that and it's still running. 271 00:25:50,020 --> 00:25:53,020 So the the UK one is running until the end of this march, right? 272 00:25:53,530 --> 00:25:58,749 By which time we'll have done three years and that'll be enough. I mean it's great. 273 00:25:58,750 --> 00:26:05,950 But when, when you have to look at the data, I've looked at the data every evening for the last three years without exception, 274 00:26:06,430 --> 00:26:09,760 because I'm the one that keeps whether the data is good enough to go or not. 275 00:26:10,340 --> 00:26:15,280 Um, and it's fine. We've, we've got it down to find out. And I can tell within 10 minutes whether we got problems or not. 276 00:26:15,610 --> 00:26:19,030 But if there are problems that I need to be calling up the lab managers and saying, Well, 277 00:26:19,600 --> 00:26:23,620 this play didn't run very well, do we understand why are the robots behaving properly? 278 00:26:23,920 --> 00:26:28,090 Have we got issues with reagents? There's a lot of moving parts that can go wrong on a daily basis. 279 00:26:29,080 --> 00:26:34,390 So yes, I shall not miss that. And does that mean and is that going to stop collecting from that household survey? 280 00:26:34,440 --> 00:26:38,110 Correct. Correct. Right. There's no more funding. There's no there's no more funding. 281 00:26:38,120 --> 00:26:49,150 Yeah. Oh, yeah. Interesting. So the other thing that I did during the pandemic was to support a really unusual project called Combat, 282 00:26:49,870 --> 00:26:54,909 um, which is basically what is called a multi modal blood atlas. 283 00:26:54,910 --> 00:26:58,170 Not only yes or multimodal in this case I think. 284 00:27:00,070 --> 00:27:08,020 Yeah, but I'll explain why I call it multimodal in a minute but yes multi I make it and the idea behind this was that 285 00:27:08,110 --> 00:27:17,670 there was already a recognition really early on that there may be some similarities between the way COVID um, 286 00:27:18,760 --> 00:27:25,150 has an effect upon your, your respiratory pathway, um, and flu and sepsis similarly. 287 00:27:25,540 --> 00:27:34,960 And there was a wonder whether there might be really common reasons why these things are happening between these different diseases effectively. 288 00:27:35,560 --> 00:27:40,420 And so what was remarkable this project was um, well, a number of things were remark about it. 289 00:27:41,200 --> 00:27:49,059 Remarkably, we had access to around about 150 patients with flu, sepsis and COVID and COVID, different levels of severity, including unfortunately, 290 00:27:49,060 --> 00:27:56,920 those who died and were able to get blood samples longitudinally at different time points during their disease progression. 291 00:27:57,430 --> 00:28:06,190 And um, we were then able to amass a lot of expertise and capability across the university to be able 292 00:28:06,190 --> 00:28:13,000 to use those blood samples on different pieces of kit to ask different sorts of questions. 293 00:28:14,520 --> 00:28:25,950 Carry on. Yes. So we have the blood samples that we were able to run across different types of technologies to understand what features 294 00:28:25,950 --> 00:28:32,250 there might be in the blood samples that would describe the different severity levels of the COVID or the flu or the sepsis. 295 00:28:32,940 --> 00:28:36,270 And why this was remarkable was a couple of reasons. 296 00:28:36,300 --> 00:28:40,650 First of all, you know, everybody wanted to help out in the COVID context. 297 00:28:40,650 --> 00:28:48,630 So we had I think it was over 120 researchers from 20 different departments across three different divisions in the university. 298 00:28:49,170 --> 00:28:55,350 That has never happened before. And we were all able to work together because none of us were in, while very few of us were in. 299 00:28:55,350 --> 00:28:58,530 Right. And then I think I'll talk maybe talk more about this later. 300 00:28:58,530 --> 00:29:07,049 This was the beginnings of a really strong recognition of the way collaboration can occur without walls going forward. 301 00:29:07,050 --> 00:29:15,150 So that, that, that was very cool. Um, I think the other cool thing was that we made a decision early on that, um, yes, the, 302 00:29:15,270 --> 00:29:22,530 the pieces of kit, the laboratory platforms were in different platforms around the university and, 303 00:29:22,530 --> 00:29:27,450 you know, we'd have to get the blood samples to them, they would run them there, they would get the data, but the data wouldn't sit there. 304 00:29:27,720 --> 00:29:30,870 And so the data would be put into a central data warehouse, 305 00:29:31,200 --> 00:29:36,180 which everyone would have access to and everyone would be able to operate upon in the same place. 306 00:29:36,600 --> 00:29:45,360 And the beauty about this is that we didn't have we avoided the usual problems of people not knowing what the gold standard data is. 307 00:29:45,360 --> 00:29:51,899 And is this a the copies of the right version? So we had to put things in like identifiers for particular data releases. 308 00:29:51,900 --> 00:29:57,000 So, you know, we made clear that if anyone presented something within the combat team that was used, 309 00:29:57,000 --> 00:30:00,300 the identifiers, everybody knew what they were talking about and how they got there. 310 00:30:00,810 --> 00:30:07,770 So the provenance was well captured, um, that it was structured in a way that we understood who and what was structured in 311 00:30:07,770 --> 00:30:12,480 a way that we knew what had happened to that data and that really paid dividends. 312 00:30:12,930 --> 00:30:16,620 And that was a new way of working for many people, something I've done for a long time. 313 00:30:16,620 --> 00:30:23,010 But across so many departments and so many people that was quite revolutionary here in Oxford, 314 00:30:23,980 --> 00:30:30,720 it might be more normal now that universities where it's it's a little more top down, but in a federated place like this, that's very challenging. 315 00:30:31,170 --> 00:30:37,379 So that was pretty cool. Um, and being able to take the clinical data, 316 00:30:37,380 --> 00:30:45,120 the clinical metadata and be able to integrate that with molecular and cellular data and together, that's what we call multimodal data. 317 00:30:45,630 --> 00:30:49,410 It includes the MULTI-OMICS, but it also includes the clinical data. And that's why it's multi-modal. 318 00:30:51,060 --> 00:30:53,430 That really hasn't been done in this way before either. 319 00:30:53,430 --> 00:31:00,030 So in that the people who are looking at the particularly the genomic data were able to stratify 320 00:31:00,030 --> 00:31:05,280 that what they were seeing relative to what the clinicians had marked down by the sample. 321 00:31:05,460 --> 00:31:09,210 Normally there's a big chasm between and you know, you can't do that dynamically. 322 00:31:09,660 --> 00:31:16,500 So that was very cool. And of course what you're saying is that somebody say someone who was especially sick with COVID 323 00:31:16,500 --> 00:31:21,330 or a group of people who seem to be especially sick might have a particular genetic signature, 324 00:31:21,390 --> 00:31:25,920 for example. And that's that's, you know, the technical terms. 325 00:31:25,920 --> 00:31:35,010 We were looking for interesting biomarkers and yet being able to segregate what you're seeing relative to the actual symptoms of the disease. 326 00:31:35,580 --> 00:31:38,700 Clearly, I mean, it's it's it's common sense. That's the right thing to do. 327 00:31:39,000 --> 00:31:42,780 But you'd be surprised how hard it is to do that in normal practice. Yeah. 328 00:31:43,560 --> 00:31:45,000 And I don't know if you had anything to. 329 00:31:45,030 --> 00:31:52,900 I did look at their papers, had anything to do with how the data was represented graphically, because that seemed no active. 330 00:31:52,950 --> 00:31:56,339 I didn't, but I had not directly. 331 00:31:56,340 --> 00:32:06,900 But I did have some input into bringing in Steve Taylor in particular, who has, um, some really cool ways of presenting quite complex, 332 00:32:07,790 --> 00:32:16,529 multi dimensional data in a way that if you're not computational but you are a biologist, you might be able to interpret and navigate. 333 00:32:16,530 --> 00:32:23,220 And that that was pretty cool. Yeah, exactly. And how long did that? 334 00:32:23,970 --> 00:32:27,560 Oh, um, so I think we started it. 335 00:32:27,570 --> 00:32:30,810 It was kind of mooted around June-July time that we were going to do this. 336 00:32:31,260 --> 00:32:34,920 We got emergency funding from the university or the division to do it. 337 00:32:35,670 --> 00:32:41,760 Um, and things really, it really took a year to get most of the work done. 338 00:32:41,760 --> 00:32:49,480 And for the first draft of the cell paper, um, to be pulled together, I mean, just even pulling that cell paper together was a, 339 00:32:49,740 --> 00:32:56,790 a feat in itself with so many collaborators and so much data and trying to come up with the 340 00:32:56,790 --> 00:33:01,199 one story that made sense because there were so many interesting sub stories out of it. 341 00:33:01,200 --> 00:33:09,690 But what is the one story that you can sell in the cell paper, and what do you think has been the impact of of that paper? 342 00:33:10,170 --> 00:33:12,749 Not as much as I think we might have imagined. 343 00:33:12,750 --> 00:33:22,920 I think in the end, a lot of the data that we found, um, replicated many other papers, but we did it all in one go. 344 00:33:23,490 --> 00:33:30,630 So I'm not sure there's anything that was, uh, unexpected by the time we had the paper accepted. 345 00:33:31,170 --> 00:33:35,129 But when we were writing the paper, there were the papers out there that said otherwise. 346 00:33:35,130 --> 00:33:40,170 But as unfortunate as we went through the reviewing process and other things came out, so. 347 00:33:40,410 --> 00:33:44,100 So I think the impact from that perspective was a lot lower than perhaps you would have liked. 348 00:33:44,580 --> 00:33:50,000 But I think the biggest impact has been, as I mentioned earlier, the, um, 349 00:33:50,280 --> 00:33:56,159 the change in perspective about how collaborations should be done and the recognition that 350 00:33:56,160 --> 00:34:00,650 it is possible to work across departments and it is possible to be less insular and, 351 00:34:00,810 --> 00:34:09,510 and it is almost expected that you have a multi skilled, multi, you know, talented approach to solving a particular problem. 352 00:34:10,020 --> 00:34:13,110 Um, and so that's one aspect. And the other aspect is the data management. 353 00:34:13,110 --> 00:34:20,820 Now everybody wants to do what we did and so I'm very fortunate that we're in really high demand to replicate this on that project by project basis. 354 00:34:21,270 --> 00:34:28,110 I was definitely including that in yes, I was thinking of talking about this as in so so that's giving you a lot more work to do. 355 00:34:28,410 --> 00:34:36,120 Oh, yes. So I just wanted to pick up on that collaboration point again, because I think it's very important. 356 00:34:36,960 --> 00:34:45,360 I mean, even even within labs and certainly between institutions, academic life can be quite competitive. 357 00:34:46,110 --> 00:34:48,059 But clearly what was going on during COVID, 358 00:34:48,060 --> 00:34:54,000 a lot of that fell away because of the urgency of putting minds together to crack, but not just the urgency. 359 00:34:54,270 --> 00:35:00,030 I mean, the that there was no competition because none of this was anybody's field. 360 00:35:00,960 --> 00:35:07,560 So this and everybody knew that if they didn't play, it was going to have a negative impact upon their career anyway. 361 00:35:07,770 --> 00:35:11,099 So a lot of the boundaries broke down as a consequence, 362 00:35:11,100 --> 00:35:19,560 and many of them have stayed down because many of my colleagues have recognised that this is they might not felt comfortable doing it this way, 363 00:35:19,650 --> 00:35:25,530 but it worked really well and is now expected and there are always going to be some that don't see that, 364 00:35:25,530 --> 00:35:27,959 but they're going to fall to the wayside because this is just the new way of 365 00:35:27,960 --> 00:35:34,980 working and it's fine to have to be one name among 200 000 people within reason. 366 00:35:35,000 --> 00:35:38,070 Well, for something like that, yes, absolutely. 367 00:35:38,850 --> 00:35:41,730 But the pressure then is to get it in the very best, general, right? Yeah. 368 00:35:42,000 --> 00:35:50,820 Um, but the data that we generated in combat and even the data we generate in the serology platform has spawned so many subprojects and, 369 00:35:52,230 --> 00:35:56,850 and papers with small authorship. Everybody's benefited in some ways, somehow. 370 00:35:57,310 --> 00:36:00,330 Mm hmm. And so what are you mainly working on now? 371 00:36:01,050 --> 00:36:11,640 Well, uh, so my Sandy research informatics team is busy providing lots of data management solutions to collaborators around here, 372 00:36:11,640 --> 00:36:16,200 collaborators around the world, particularly around the early, early stage drug discovery aspects. 373 00:36:16,770 --> 00:36:21,900 Um, I've just joined the welcome Centre here to help set up a big new data hub platform, 374 00:36:22,350 --> 00:36:27,030 which is kind of the kind of like that the son of the it's a big thing that we did in combat. 375 00:36:27,540 --> 00:36:34,020 Um, I and I work very closely at a divisional level to make sure that we're doing the right thing around digital information as well. 376 00:36:34,290 --> 00:36:41,669 Mm hmm. And so I think we've we've got there very fast that you'll tell me if we've left 377 00:36:41,670 --> 00:36:47,010 anything but the how did working through the pandemic impact on you personally? 378 00:36:47,010 --> 00:36:51,989 Um, you mentioned one point that none of you were in. Um, well, that was the least of our problems. 379 00:36:51,990 --> 00:37:04,500 I mean, it was, it was really interesting how course many colleagues who could not find things to do during pandemic, um, really were bored silly. 380 00:37:05,460 --> 00:37:11,640 And, you know, there's only so many papers you can write and papers you can read and but when you can't be in a lab and 381 00:37:11,640 --> 00:37:18,210 you can't generate data and this is very damaging not only to their careers but also their mental health. 382 00:37:19,020 --> 00:37:28,680 Um, I, you know, I had students who. Could not be in the lab and need to be I students who were purely computational and therefore could work at home. 383 00:37:29,220 --> 00:37:32,610 But yet they were stuck here in Oxford, isolated. 384 00:37:32,700 --> 00:37:39,600 They couldn't be around friends, they couldn't be around family. And, you know, one of them basically lost a year as a consequence of that, 385 00:37:40,350 --> 00:37:44,639 forced the university, put in some fairly good safeguards around that and we could mitigate it. 386 00:37:44,640 --> 00:37:48,270 But nevertheless, that that had a huge impact upon people. 387 00:37:48,540 --> 00:37:54,200 For those of us that were working at kind of like the coalface, we never worked so hard in our entire lives. 388 00:37:54,720 --> 00:37:56,010 And it was, you know, 389 00:37:56,010 --> 00:38:02,550 there were some days when I was up to two or three in the morning trying to unravel why the data from the previous day wasn't right. 390 00:38:03,600 --> 00:38:07,470 And then, you know, you you have an 8:00 meeting every morning just to summarise where you are. 391 00:38:07,710 --> 00:38:11,010 There wasn't an awful lot of sleep and that had a major impact upon my family. 392 00:38:11,220 --> 00:38:16,550 Bluntly, they didn't see much of me and that was tough for them too, I'm sure. 393 00:38:16,850 --> 00:38:26,430 I'm sure. And was there anything you were able to do to support your, your wellbeing through that, or was that just not time? 394 00:38:26,610 --> 00:38:31,469 This wasn't time. I mean, to be clear, I mean, we weren't as busy as the the people in the. 395 00:38:31,470 --> 00:38:36,209 Jenna, right? I mean, they really were I mean, they made us look like we were hardly working at all, 396 00:38:36,210 --> 00:38:41,070 But, um, you know, it was just constant and it was even at weekends, 397 00:38:41,380 --> 00:38:45,300 phone calls and, you know, something's broken or we need to think about this, 398 00:38:45,300 --> 00:38:48,840 or the government's asked for this and they want it with the next 3 hours. 399 00:38:48,840 --> 00:38:52,820 You know, it's tough when they things like that happen, so you just have to plough on. 400 00:38:52,830 --> 00:38:56,160 But, you know, I was at home, right? So I wasn't coming in. 401 00:38:56,700 --> 00:39:02,759 So the family saw me, but they kind of didn't see me. And how threatened did you feel by the virus itself? 402 00:39:02,760 --> 00:39:07,380 By the possibility of infection? I assumed I was going to get it. There was no doubt about it. 403 00:39:09,090 --> 00:39:12,120 But I was concerned about that because my wife is in a high risk category. 404 00:39:12,690 --> 00:39:16,320 Um, actually, to this date, none of us have had it. 405 00:39:16,920 --> 00:39:20,970 I hope it stays that way. But yeah, it was always a concern. 406 00:39:21,540 --> 00:39:26,339 Yeah, but you were. Yeah. You were staying at home. 407 00:39:26,340 --> 00:39:31,720 You weren't, actually. Well, I mean, one of the things I'm responsible for is a lot of i.t. 408 00:39:31,740 --> 00:39:35,070 Infrastructure, um, which support was supporting what we're doing. 409 00:39:35,490 --> 00:39:38,700 So I did have to come in occasionally, but of course the place was deserted anyway, 410 00:39:39,180 --> 00:39:43,829 and you were just scrupulously careful about what you touched and how you did things. 411 00:39:43,830 --> 00:39:47,010 And it was fine. It worked okay. Yeah. 412 00:39:48,120 --> 00:39:53,669 And what I didn't pick up earlier, you mentioned the government was one of the things you were dealing with. 413 00:39:53,670 --> 00:39:57,570 Was that something you, you were you personally interacting with civil servants? 414 00:39:57,780 --> 00:40:01,649 Yes, yes, yes. Because that was a new experience for you, Of course. 415 00:40:01,650 --> 00:40:10,229 Yes. Um, but I think, you know, and in normal times, I think it would have been harder, but because we were all in the same boat, you know, 416 00:40:10,230 --> 00:40:19,530 we recognise that civil servants were under enormous pressure by the Government to deliver answers and to deliver data to be able 417 00:40:19,530 --> 00:40:25,230 to explain to the population what was going on and why it was important that they were making the decisions they were making. 418 00:40:25,830 --> 00:40:32,310 And, you know, it wasn't unusual for, you know, to find yourself on a call with civil servants very, very stressed out, taking it out on you. 419 00:40:32,730 --> 00:40:38,220 And you know, you completely understand that and you find ways to work collaboratively with them, 420 00:40:38,610 --> 00:40:41,669 um, to help them out, because that's, that's how it worked. 421 00:40:41,670 --> 00:40:47,760 And in the end, I mean, there was quite a lot of turnover of civil servants over that time because I think a lot of them found it really hard. 422 00:40:48,210 --> 00:40:52,680 Um, but we had lots of very good working relationships with them and, and you know, it was good. 423 00:40:54,000 --> 00:40:59,050 And did the fact that you were working on something that was so important to you, 424 00:40:59,100 --> 00:41:02,159 do you think, that supported you or your own wellbeing through this very stress? 425 00:41:02,160 --> 00:41:04,739 I don't think I didn't realise when we first started. 426 00:41:04,740 --> 00:41:14,190 I didn't realise how important it was and actually it really wasn't until it really wasn't until probably end of 2020, early 21, 427 00:41:14,190 --> 00:41:20,040 when it became clear when the big national lockdowns were occurring and the numbers were appearing, 428 00:41:20,040 --> 00:41:23,790 you know, in the Downing Street, it was only then that I realised I hang on, that's my data. 429 00:41:25,260 --> 00:41:30,209 And because there's so many steps removed from, you know, you producing the data, 430 00:41:30,210 --> 00:41:35,340 giving it to an S own s does the analysis, then they give it to DHS agency, then talk to the civil servants. 431 00:41:35,340 --> 00:41:38,639 You then talk to number ten and you don't see it. 432 00:41:38,640 --> 00:41:45,000 Right. And, and actually we on the serology platform really didn't understand what happened to the data afterwards. 433 00:41:45,000 --> 00:41:48,180 And we weren't about to go and ask go and ask because they were stressed out enough as it was. 434 00:41:48,570 --> 00:41:53,390 So it wasn't really until the middle of 21 end of 21 that we were able to get a 435 00:41:53,410 --> 00:41:57,720 sit down and actually show us how were they using the data in their modelling. And it's like, Oh, that's very cool. 436 00:41:59,040 --> 00:42:02,190 But at the time we didn't know how important it was. 437 00:42:02,970 --> 00:42:08,940 Then it seemed like far more important things going on with the world just felt like we were just a small thing that Matt might be helping. 438 00:42:10,360 --> 00:42:19,970 Oh, Um, and are there any, any particular stories that stick in your mind from the time that, um. 439 00:42:22,120 --> 00:42:28,239 I'm not good at remembering these things. Not really. 440 00:42:28,240 --> 00:42:33,760 I think it was just trying to make sure that we could deliver, 441 00:42:34,360 --> 00:42:46,480 trying to look after people and people's mental health on the team because they were going above and beyond even those on shift work. 442 00:42:47,170 --> 00:42:49,840 You know, they were staying longer to make sure that things were done. 443 00:42:50,950 --> 00:42:54,730 They were getting very stressed when things weren't working, which was good but bad. 444 00:42:55,330 --> 00:43:00,309 Um, and just the way that, you know, if, if there was a problem, 445 00:43:00,310 --> 00:43:04,300 we all kind of mucked in together and made it happen and we didn't always get the right solution, 446 00:43:05,230 --> 00:43:10,780 but we got to the point where the data was always acceptable to those who cared. 447 00:43:16,490 --> 00:43:19,610 So I think. I think I've reached the end. Oh, there you go. 448 00:43:20,420 --> 00:43:25,190 Which is in the last question is really whether the experience of working through 449 00:43:25,190 --> 00:43:30,440 those code projects has changed your attitude or your approach to Absolutely. 450 00:43:30,860 --> 00:43:33,200 100%. Um, in what way? 451 00:43:33,450 --> 00:43:42,650 In the way that I can now have confidence that if someone throws something big at me, there's a way in which we can get it done. 452 00:43:43,190 --> 00:43:46,220 And it's not about me. It's about the team and the collective. 453 00:43:47,540 --> 00:43:54,290 And is there anything you'd like to see change in the future with regard to how things are organised? 454 00:43:55,310 --> 00:44:00,170 Well, I mean, I mean, I mentioned the aspects of collaboration and, you know, 455 00:44:02,150 --> 00:44:06,050 the university set up in a particular way that has served it very well over a long period of time. 456 00:44:06,470 --> 00:44:14,570 I think there is a lot of understanding that things change are changing and I don't think there's a good solution to this, 457 00:44:14,570 --> 00:44:23,299 but we have to find approaches and methods that support that collaborative way of working in a more seamless 458 00:44:23,300 --> 00:44:30,380 way and to lower the energy barrier to catalyse sort of research that we can do if we work together. 459 00:44:32,390 --> 00:44:33,680 Thanks very much. That's right.