Episode 49 • 13 June 2022

Ajay Karpur on Metagenomic Sequencing

Leave feedback ↗
Contents

Ajay Karpur is an engineer and entrepreneur working on projects related to biosecurity. He’s hoping to start tweeting again soon, at @ajaykarpur.

Joining as a guest co-host on this episode was Janvi Ahuja, who is a PhD student in computational biology at Oxford University, an affiliate with the Future of Humanity Institute, and since recording this interview, part of the Johns Hopkins Centre for Health Security ‘Emerging Leaders in Biosecurity’ program. She’s tweeting at @jn_ahuja

(Image) Ajay Karpur

In our conversation, we discuss:

Ajay’s recommendations

Resources

On disaster preparation

For prospective biosecurity engineers

Transcript

Intro

Fin 0:06

[…] Our main focus in this conversation was a technology called ‘metagenomic sequencing’, which is a way of comprehensively looking at all the genetic material in some sample, rather than just looking for a specific known pathogen, like how PCR tests for COVID are just looking for COVID but will ignore other things. Now, when experts try to imagine a world that is secure against future pandemics, including pandemics worse than COVID, it turns out that having this ability to really quickly and cheaply spot new pathogens just looks hugely important. So we talk about how to make sure this technology becomes much more accessible, and how metagenomic sequencing could make the world safer against biological threats. We also talk about new ways of funding high impact research, refuges against worst case pandemics, and how and why you should get involved with this work if you have an engineering background. Big thanks to Ajay and Janvi. I learned so much on this one. Okay, here’s the episode. Ajay and Janvi, thanks very much for coming on.

Ajay

Thanks for having me.

Janvi

Yeah!

Fin

So one question we’ve started asking everyone Ajay is: what is a question that’s just on your mind right now that you’re kind of stuck on?

Ajay 1:54

Hmm. That is a great question. I guess one of the big things I’m thinking about right now is what are all of the technology bottlenecks to making progress in metagenomic sequencing? So I guess, how can we take the technology from where it is today, to where it needs to be by the end of the decade, so we can build an early warning system for emerging pathogens.

Fin 2:18

Sweet. Well, let’s talk about that then.

Ajay

Sure.

What is metagenomic sequencing

Fin

So I’ve been hearing about metagenomic sequencing a lot more recently. I’m an outsider to biotech stuff; can you explain just in very general terms, what is metagenomic sequencing?

Ajay 2:32

Yeah. So genomic sequencing is the reading of genetic material from a sample, and metagenomic sequencing is, I guess, at a very high level, reading all of the genetic information from a given sample. So typically, what’s done is something called ‘amplicon sequencing’, or some variant of that, which just means you take a primer. So this could be like a genetic substring, I guess, is one way of thinking about it. And you’re saying, I want to amplify everything in my sample that has this substring in it. And what that allows you to do is effectively enrich your sample. But the downside is that you don’t see all of the things that you didn’t enrich. So the key with meta genomic sequencing is that it’s pathogen agnostic. Or in the literature, this is sometimes referred to as ‘unbiased, hypothesis free’. The reason that’s important is because if we want to be able to detect novel pathogens that are emerging, we need to be able to see the things that we weren’t looking for.

Fin 3:42

Got it. So to try saying that back: I’ve got my sample of, you know, saliva or whatever, and typically, in genomic sequencing, the specific test I’m using already has in mind a pathogen which it’s testing for, and there’s some procedure by which it kind of amplifies that pathogen, but kind of throws away or ignores everything else. So it’s really good at checking whether in fact I have such and such a disease, but it’s not going to look at everything in that sample. And in fact, that sample is going to be full of all other kinds of things as well.

Ajay 4:17

Yeah, that’s right. So tools like PCR culture, serology, amplicon sequencing, and, and even CRISPR diagnostics don’t satisfy this key constraint. They’re not pathogen agnostic. There’s this kind of funny story or a joke: there’s a man who’s lost his keys. And it’s at night, he’s kind of crawling around on the ground trying to find his keys underneath a lamppost, and a police officer comes by and asks, ‘What are you doing?’ And he responds, ‘I’m looking for my keys’. The police officer asks, ‘Did you lose your keys right here under the lamppost?’ and the man responds, ‘No, but this is where the light is’, and that’s, you know, effectively what we’re doing with existing diagnostics. And what we need to be able to do is look at everything that exists in a sample.

Fin 5:08

Cool. Got it. And I guess one question is why? Why is it especially important or useful? Maybe it’s just totally obvious, but can you just speak to why it might end up mattering a lot that we can detect novel pathogens, as well as all the ones we currently know about?

Ajay 5:25

Yeah. So the hope is that we can have this capability deployed ubiquitously. And when we have that kind of deployment, then when something new emerges, that’s potentially a very high consequence pathogen - it could be highly transmissible, highly virulent, have other properties that make it concerning - we want to be able to detect it as early as possible so that we can act to contain it. And should it not be containable, we want to provide as much lead time as possible, so that other measures can be taken.

Fin 6:02

Got it. Cool. I guess one way to spell this out is just to ask you to maybe try describing a world in x decades from now, in which metagenomic sequencing is cheap enough to be close to ubiquitous - what does that look like, when does it get used? That kind of thing. Yeah, so

Ajay 6:21

I guess there are three general categories of contexts where you might see it deployed. Those are clinical, sentinel, and environmental. So clinical is what you’d expect: maybe you could walk into - so I guess there’s what might be possible a decade from now, and then there’s what might be possible, you know, many decades from now -

Fin

Let’s go for many decades from now and work back.

Ajay

Yeah, so I guess a fourth context would be like point of person or point of home. So in the far, far future, you might imagine that you wake up in the morning and go to the bathroom. And before you brush your teeth, you just spit into a tube that, you know, scans and tells you whether you have any of a number of known pathogens. And if there’s something that it’s never been seen before, we’ll look at the sequence, predict something about what its functions are and assess whether it might be concerning.

Fin

Okay, cool.

Ajay

Yeah, but I guess, maybe nearer term, the other three contexts are a little more relevant. So you can imagine in a clinical context, let’s say you have a cold or something like that, or you think you do, and you go to the doctor, and ask for a diagnosis. And the doctor can just take a sample from you - just a, you know, crude liquid sample, could be saliva, loaded into a cartridge - stick that cartridge in a box that they have sitting on their countertop, and then press a button, and then less than an hour later, they’ll have a list of all of the pathogens that were in that sample, whether they were things that they’d seen before or not. And then if it’s something that has never been seen before, and is potentially concerning, then that information could be networked together with other devices that are also providing this kind of sensing capability, providing like this live data stream to public health authorities.

Fin 8:26

Does this have something to do with a sentinel idea you mentioned?

Ajay 8:29

Yeah, so sentinel would be the idea of regularly sampling from people who are likely to be at higher risk of exposure to pathogens. So these could be infectious disease researchers, they could be, you know, TSA agents, who are coming into contact with people on a regular basis, they could be people working at high density livestock facilities. Anywhere where these people serve as kind of an early warning for the spread of a pathogen.

Fin 9:04

Okay. So they’re most likely to be exposed earliest, and therefore, they’re good people to focus on, if you want to get the message out as soon as possible.

Ajay 9:12

That’s the idea. Yeah. And also you want to be able to detect things that are spreading asymptomatically as well. And if you’re only looking at people in the clinical context, you’re only going to be getting samples from people who are symptomatic - for some reason I’m coming into the doctor.

Fin 9:40

Yeah, got it. So I guess one thing I’m wondering is: in either case, whether I’m symptomatic or not, maybe I give you my sample, you throw it into the machine, and - tell me if I’m wrong - but presumably, it’s going to pick up a bunch of known pathogens, and then a bunch of unknown pathogens, but maybe only a small subset of those unknown pathogens are actually really worth worrying about. Is there anything that we can do to kind of get a sense of which of these unknown pathogens are most likely to be the serious ones?

Ajay 10:09

Yeah. So there are a number of tools that allow you to take a genetic sequence and predict its function. And there are a couple of projects out there right now. For example, the Secure DNA project out of Kevin Esvelt’s lab, working on providing this kind of predictive capability in a safe and secure way.

Fin 10:34

Sweet. And that’s just a computational challenge, presumably?

Ajay 10:37

Yeah. There are existing bioinformatics tools, as well as emerging methods that can allow for that kind of triage capability.

Janvi 10:45

Can we maybe go through what the different types of surveillance look like? So what would environmental surveillance look like, and clinical and then sentinel look like? And it also seems like when we’re talking about a world that’s tens of years ahead, and everything is super ubiquitous, we maybe wouldn’t need sentinels so much, just because everyone is constantly being sequenced. So what are the types of sequencing that we want, like, much sooner, or in the next 10 years? And then what are the types of sequencing that we’d want in this more perfect world, tens of years from now?

Ajay 11:20

Yeah, I think that’s right. I think sentinel testing is more important earlier than it is than it is later, but I think it’s probably still going to be an important component down the line. It is also worth noting, for anyone who’s not already familiar, that we’re talking about pathogen sequencing, not constantly sequencing human DNA at all times, both for privacy reasons and also for feasibility reasons. In a given sample, there’s a lot of human DNA, and it makes it hard to see everything else that’s in the sample. So I guess, clinical, I kind of already described that a little bit, where we want anyone who’s showing up at a clinic - could be urgent care, could be hospital, could even be, you know, eventually primary care, should we drive down the cost and turnaround time sufficiently that that’s feasible. And not only does that provide a useful diagnosis to the care provider, but it can also serve as this early warning for anyone who’s subscribing to that data feed of pathogens, location, time, unique identifier, you know, potential threat level. And so that’s the clinical context. Sentinel is similar to clinical in that it’s drawing from human samples. So very likely, the types of devices that we’d be looking at there are fairly similar, in that we’re taking a crude liquid clinical sample from humans, running that through a device and getting an answer. Environmental looks fairly different. There’s a number of different contexts in which sequencing could be deployed. For environmental early warning, there’s wastewater, waterways, air in, for example, train stations or airports, you can sample from the air, sample from the wastewater systems. It’s also, I guess, not necessarily the case that sequencing is the only way to go for environmental; you could imagine things like Raman spectroscopy also being useful, where you have these like really large volumes of fluids or other samples that you want to collect, and you just want to be able to answer: is there something new here, and is it spreading? Or is it growing exponentially in my sample? And then once you have that information, then maybe you can go back and do some more extensive sequencing. So these are, you know, a lot of open questions, and the folks working at the Nucleic Acid Observatory can talk a lot more about environmental sequencing.

Fin 14:18

Cool. So broadly speaking, it sounds like there are three kinds of use cases for metagenomic sequencing, once we get it, and they are pretty cool. So I come into the doctor and maybe as part of my visit, I can take a sample; there’s this kind of sentinel idea where if I’m likely to be exposed early to a novel pathogen, then that might be a reason to like, get some testing going on there; and then thirdly, atmospheric sequencing, which is like you mentioned, maybe in wastewater or even air. I guess one question just to get this concrete in my head is: if you imagined that we had metagenomic sequencing in 2019, how might things have gone differently?

Ajay 15:03

You know, it’s actually interesting. Metagenomic sequencing was used to identify SARS-CoV-2, the issue is that it was just circulating for a long time before that happened. I think, if I remember correctly, some estimates are about 45 days between emergence in humans and actual detection. And what that meant is that you have this large population of people who have this pneumonia of unknown origin. And eventually enough of them showed up at clinics that people thought it was prudent to figure out what exactly was going on. And so they then, you know, cultured the samples, isolated and identified SARS-CoV-2. And I guess maybe one reason that that took so long is that metagenomic sequencing is this kind of tool of last resort, as opposed to this frontline diagnostic. And in order to get it to the point where it’s ubiquitous enough that, you know, maybe within like the first couple of days of the emergence in humans it can be detected, we need to be able to drive down the cost and turnaround time. So it’s feasible as a frontline diagnostic.

Fin 16:18

And what is a reasonable price of just getting a sequence done right now?

Ajay 16:25

So for metagenomic sequencing, the estimate for cost of goods per sample is probably pretty close to about $5,000. So it’s quite expensive. And then the fixed cost of the instrument you actually use is obviously substantially more expensive than that. And what we want to be able to do is drive it down to the point where it can compete with where PCR is today. So you can imagine something close to $1 cost of goods sold per sample, and, you know, $500, to, you know, a couple of $1,000 for the instrument. So there’s a lot to be done to figure out how we can accomplish that.

Fin 17:14

Yeah. Okay, maybe this is a dumb question, but what are the comparable costs for PCR? Like how much is a big PCR machine going to cost me?

Ajay 17:23

Yeah. So I guess it depends on how large the machine is and whether it’s at, you know, a very automated kind of centralised lab facility, as opposed to something that’s a little bit more distributed, but anywhere from, you know, a couple of $1000s. I guess at the low end, there are actually kinds of DIY devices that you can build for like a few $100, I think, like $500, or $600, or so. And then most are in the few $1,000 range, like, you know, $2,000 to $5,000. And then for the very large kind of multiplex devices, those can be substantially more, you know, $10,000 - $20,000.

Roadmap for making sequencing techniques financially viable

Fin 18:03

Yeah. So if I’m getting this right, you’ve been thinking recently about how we can actually begin driving down the costs and thinking systematically about what are the major blockers and the kind of roadmap from here today. So yeah, I’m curious to hear about first of all, what does that process look like? And then eventually, we could talk about what things have you actually identified?

Ajay 18:26

Yeah, definitely. So as you mentioned, we’re kind of going through this technology roadmapping process where we look at, I guess, the landscape of current technologies used in every step of the process, from sample acquisition, to sample preparation, to sequencing, to analysis, and then also reasoning from first principles to understand what the constraints are. So I guess there’s a number of different constraints; there’s the regulatory constraints, market constraints, and then there’s like the hard physics, space constraints. And within the physics, space constraints, we want to try to identify what are the true bottlenecks that are preventing, you know, order of magnitude improvements on some of the metrics that we care about. And then from there, the idea is to come up with workarounds for some of these bottlenecks that can potentially, you know, lend themselves to research and development projects or some other means of kind of attacking that bottleneck.

Fin 19:36

Yeah. So what’s the idea here? Is it that some of these bottlenecks just look really hard to kind of crack head on and so we need to spot these in advance and start thinking about alternative workarounds or something?

Ajay 19:51

Yeah. I guess the idea here is that we want to accelerate the research and development of some of the subcomponents of the technology that can, if progress is made, you know, for example, in microfluidics, or in, you know, nanopore sequencing or optical sequencing, or one of these other kind of pieces of technology that allows for the whole thing end to end to become faster and cheaper, then, you know, we want to do whatever we can to make progress there.

Fin 20:27

Cool. And are there any major blockers where it’s like, you know, this component or this process, this is just the key thing that we still haven’t quite figured out?

Ajay 20:38

I think we’re still working to identify what the true bottlenecks are. There are, I guess, some higher order things that, obviously, are blockers. There’s the cost that I mentioned, the turnaround time, there’s the complexity as well. It requires, you know, skilled labour to be able to take a crude sample, and then prepare it and actually sequence it and do something useful with it. And ideally, what we want is to just take that crude sample, as mentioned, stick it in a cartridge, put it in a box and press a button and then walk away. It should take less than a minute to accomplish that, and should be able to be done by someone who has very little or no training. But I guess at the more fundamental level, the constraints that we’re trying to understand are pretty complex, and there are a lot of interdependencies between them. So there’s a lot of work still to be done to kind of disentangle each of the each of those bottlenecks.

Janvi 21:43

And is there a reason this kind of work hasn’t been done before?

Ajay 21:47

You know, it’s a good question. I think probably some of this work has been done in the R&D labs of some of these larger sequencing companies. For example, Oxford Nanopore has developed the Voltrax system for automated sample preparation. I know there have been folks who’ve been thinking about completely integrated devices. I guess the problem we’re trying to solve, though, is one that not a lot of sequencing companies are tackling. Most of the sequencing industry is driven by oncology, and it also assumes that you have some sort of centralised sequencing operation, and so those kinds of constraints drive technology development in a slightly different direction. Whereas what we’re thinking about is ubiquitous, highly distributed sequencing for pathogens, which is a slightly different use case, and has potentially different implications for what pieces of the technology need to be accelerated.

Janvi 23:00

So it sounds like the way that maybe you’re approaching this, or were approaching this, is from an angle which involves different parts of the problem than individual sequencing companies would want because our overall goal is different. Does that seem correct?

Ajay 23:15

I think that’s right. Yeah, I think trying to integrate everything end to end is one way that our approaches may be different, as well as the actual, like location of deployment and the purpose of the sequencing as well.

Janvi 23:33

Does this become super difficult, because different parts of the problem are at different stages of being close to completion?

Ajay 23:41

I think that’s right. Yeah, I think there are some things that seem to work pretty well, like DNA, RNA purification, and then other things that, you know, there’s still a lot of progress to be made.

Incentives: how do nonprofits help?

Fin 23:55

Yeah, I guess the thing I’m wondering is: if I make, you know, medical devices, and I come up with the tech which drives down the cost of sequencing, well, there’s gonna be a tonne of buyers. Like, there’s a huge reason for me to try to do that. I take it when you’re doing your roadmapping research, you’re not doing that, because you have dreams of making a tonne of money, you’re doing it because it just seems really important.

Ajay

Yeah.

Fin

So why do these kinds of nonprofit interests need to step in and do this job? Why isn’t it just being fully saturated by the companies that could make money off it?

Ajay 24:37

Yeah. I guess my take is that there’s this sort of valley of death between research in emerging technologies and then things that are actually commercializable, and taking things from a low technology readiness level out to things that are mature enough to be deployed commercially is very difficult and many techniques don’t actually make it past that gap. So going from the lab to the market is, I think, a lot more difficult than maybe a lot of people appreciate it.

Fin 25:12

Yeah. So I guess you might break that down - I’m kind of thinking of different ways that could be true. So one reason might be uncertainty; so if we’re not totally confident that with this huge R&D spend, we’ll actually get the thing out the other end, then we might back away from doing it. Now, there might be times we maybe just expect it to take a very long time, and it’s very hard to think over, or it’s hard to plan over long timescales. And then a third reason might be a kind of freerider-y thing, where if you expect, you know, a competitor to do the research, then maybe it’s in your interest to wait for them to do it. But they’re thinking the same thing. Which one of those sounds most accurate? Or am I missing some extra thing?

Ajay 26:03

Yeah, I think it’s maybe a bit of all the above, I think existing organisational structures aren’t really incentivized to solve this problem. And it’s a little bit more of this ad hoc process as to what makes it through. But in our case, we care quite a bit about this differential technology development. We see this emerging technology, and we know that we want this to be deployed ubiquitously. And so in this case, it makes sense to try to deploy funding to make substantial progress on the technology. And I think the reason we care about driving down the cost and turnaround time, so that it’s actually something that’s commercially viable, is just that it wouldn’t be feasible for a ubiquitous sequencing system to be completely philanthropically funded. We do need it to be driven by private markets, to some degree, or I guess, to a large degree. And I think the best place for the application of philanthropic dollars is in accelerating the research and developments so that can then be possible later.

Fin 27:17

Yeah. Can you say something about what it means to care about differential technology development? What does that mean?

Ajay 27:24

Yeah. So I guess there are some technologies that we view as inherently defensive, and others that could be viewed as potentially more offensive. And what we want to do is, I guess, on one end, deter the development of things that are purely offensive, like biological weapons, and then in the middle, for things that are dual use, maybe you want to kind of push them to the left, so that they’re more inherently defensive, or you kind of try to ensure that technologies enter the world in an order that’s kind of inherently safe. So for example, you know, control systems entering the world before nuclear weapons. And then all the way on the other end, where things are purely defensive - or at least we strongly suspect that they are - you want to accelerate the development, so that as mentioned, you know, we can get those things out to the world before they’re needed.

Fin 28:23

Got it. So you want to avoid the window in which it’s potentially too easy to do a lot of damage, and too difficult to defend against it.

Ajay

That’s right.

Fin

Yeah. Cool. That makes a tonne of sense. Was there anything you wanted to add, Janvi?

Janvi 28:41

I was just going to ask how the development of metagenomic surveillance helps with differential tech development, if you could spell that out a little bit more?

Ajay 28:51

Yeah. I guess metagenomic sequencing is a fundamental technology for defensive technology; it can be used for this early warning capability. I guess the thing we care about with early warning is both detecting and also characterising emerging pathogens. And once we’ve characterised a pathogen, we can then use that information to develop medical countermeasures. And early detection can potentially buy us the capability to contain a pathogen if we act early enough, and if the characteristics of the pathogen are not prohibitive to containment. And then should it not be containable, it buys us enough time to take other measures.

Janvi 29:40

So it seems like there could also be this third thing, which is that having this sort of ubiquitous system means that people who might want to develop dangerous things are less inclined to do so because it’s easier to get caught. Does this play into differential technology development and sort of minimising the offensive stuff that’s been developed?

Ajay 30:03

I think that’s absolutely right. There’s very much this ‘deterrence by denial’ kind of dynamic here.

Fin 30:04

You can imagine reaching a world in which the balance tilts towards defence. In other words, in some sense, you guys succeed in this differential tech development thing, in terms of biosafety. And then we don’t see any kind of crazy engineered pathogens, for instance. And some people might point to that as evidence that this wasn’t really worth worrying about in the first place. But the thing that Janvi just said suggests that that might not actually be appropriate.

Ajay 30:41

Yeah. And I think, you know, obviously, we would love to be in that world, right, where we build this system, and there aren’t any threats that mean, it’s like, strictly necessary. But I think even in that world, this system would do a lot to suppress the transmission of pathogens; if you can track their spread, and provide public health better situational awareness, so that they’re not flying blind as they were during this pandemic, and continue to be. There are pathogens all across the spectrum of, you know, how much we want to worry about them, and doing whatever we can to stop the spread of all of them is ideal. And I think if we are considering the worst case pathogens and designing for them, my suspicion is that we’ll end up doing a lot to suppress the spread of all the rest of them.

How to respond to novel pathogens

Fin 31:37

Yeah. Sounds pretty sensible to me. I guess I kind of asked about this when I asked about COVID. But one thing I’m curious about is: let’s just imagine it’s like 2050, we have fairly ubiquitous sequencing in place - so we have something like a sentinel setup. And perhaps we’re unlucky enough that a fairly dangerous, novel pathogen emerges. How does this play out? In the best case, from being spotted for the first time - what could come next?

Ajay 32:16

Yeah. So just make sure I understand, we’re imagining we have this ideal system deployed?

Fin 32:21

We have the ideal system, yeah. How does it actually react after the point at which it first picks up on something new?

Ajay 32:27

Yeah. So let’s assume that there’s this high consequence pathogen that has emerged in humans, and we have this index case, they can walk into a clinic, maybe they’re symptomatic. And it’s been only a few hours that they’ve been infectious. And when their samples are run through the diagnostic system, and the care provider has gotten the results back from the system and identified that there’s something potentially concerning here, and that they should be isolated, then I guess that’s when the reporting step kicks in. You can imagine that the diagnostic device that the sample is run through not only provides the diagnosis to the care provider, but also automates the reporting step. So if something passes a given threshold for how much we might be concerned about it, then it can automatically notify relevant public health authorities at the local, state, federal or international levels. And that’s when the risk response step kicks in. And right now, that response capability is largely the responsibility of governance. And I imagine that will be the case for the foreseeable future. But there are also companies that are working on providing not just advice to governments on response capabilities, but also providing some of those capabilities themselves.

Fin 34:09

Cool. And then in practice, what could response look like? I guess isolating is a pretty key thing?

Ajay 34:16

Yeah, that’s right. So isolating the person. Also ring vaccination is one potential if we’re kind of looking in the far future, and let’s say we have the ability to rapidly develop and deploy medical countermeasures against even given threats. So as soon as we detect it and characterise it, then you might want to vaccinate everyone who has come in contact with that person and everyone who’s come in contact with them. Kind of like a preventative measure to prevent further spread. That’s one possibility. There’s, certainly, you know, in order to determine who you might want to vaccinate, there’s some tracings you might want to do, which probably will look pretty different from traditional contact tracing.

Fin 35:10

Hang on a sec. So I don’t want to go too deep down this rabbit hole, but you did mention, ‘Oh, so we’ll just vaccinate everyone, once we spot the new pathogen’.

Ajay

Yeah.

Fin

So I thought that it takes years to develop a vaccine. So what could be going on here that allows us to create a vaccine so quickly?

Ajay 35:25

Yeah, so I guess, to clarify, not vaccinate everyone, but vaccinate those that this person has come into contact with. And so yeah, I guess we already saw that the time between, you know, getting the sequence and developing the candidate mRNA vaccine was on the order of days, for Moderna for example. And DARPA has a couple of programmes right now - one is called the ‘Pandemic Prevention Platform’, another one is called ‘Nucleic acids On-demand Worldwide’ - which are focused on this mobile, forward deployed medical countermeasure development capability where you could develop and synthesise nucleic acid MCMs, you know, anywhere in the world. And so with that capability, you can imagine that once you have a sequence, you can go from that to candidate vaccine or therapeutic in pretty short order.

Fin 36:30

Sweet. Awesome. Okay, is there anything else on the sequencing stuff from your end, Janvi, that we missed?

Janvi 36:37

I feel like something that might be worth mentioning here is kind of - to go back to the different types of sequencing we mentioned - there is this idea that we might kind of want them to exist at the same time, to provide this sort of defence, where when there are holes in some certain types of sequencing, or metagenomic sequencing, they should be picked up by others. So one way to think about this is that if we had passive environmental surveillance going on all the time, so in wastewater, we have sequencing devices that are picking up sort of any pathogen that is going on within a population. But once in a while, you know, it misses these, there’s maybe not high enough concentrations or something, we then also have clinical surveillance going on in the background. And so when people are ill, if enough people go to the hospital, and their sort of metagenomicly sequenced, depending on how far into the future we’re thinking of, then it is picked up there. And then the third one that we mentioned being sentinel - we also have these other passive systems that are looking for asymptomatic people. So the idea is kind of that we’re like covering all our bases here. And that even if some systems fail to pick up that there is like a disease of importance here, other systems do catch them.

Ajay 37:56

Yeah, I definitely would absolutely echo that. I think as much as possible we want comprehensive layer defence capabilities. And that’s layered not just in the context that they’re deployed - so environmental, clinical and sentinel - but also, as we improve capabilities, deploying the technology in phases. So you know, deploy what we have a few years from now, and then keep that operational, and then a few years later deploy what we have at that stage as well, across each of the contexts.

Janvi 38:30

Yeah. And then also in the far future, there’s a fun idea of having individuals doing their own metagenomic sequencing. One cool piece of technology that Oxford Nanopore Technologies has right now is called the ‘SmidgION’, which is this little sequencer that you attach onto your phone, and I think this is the idea of something that they want to develop. And yeah, I think having people passively testing themselves constantly is like also an idea I think we’re really excited about.

Research funding

Fin 38:59

Cool. Let’s talk a bit about research funding them. You touched on it already. It does seem pretty unusual to be doing this thing where you spot a technology, you can anticipate that there’s going to be a market for the technology, and yet, you still want to figure out how to accelerate it, not from a profit motive, but just from some kind of altruistic motive. So I kind of want to talk about weird ways of funding research a bit more, but I guess a natural first question is just: how does this kind of useful, somewhat speculative research typically get funded these days?

Ajay 39:40

Yeah. So I guess, before I answer that, one thing I want to touch on is - I think we’re actually not certain that this ubiquitous sequencing system is something that can be commercially viable, but we think the best way to accomplish that will be to drive down costs and turnaround time so that that can be possible. But yeah, there’s, I guess, a range of existing structures for high risk, high variance, sort of research and development. DARPA calls this ‘surprise’, you know, they look for technological ‘surprise’. And yeah, DARPA and the other ARPAs [Advanced Research Projects Agencies] are kind of the primary way that that kind of funding happens now. So DARPA will, you know, hire programme managers who look at technologies that depend on components that have only emerged in recent years, and put together very ambitious project portfolios that can make progress there. This is at very low technology readiness levels. There’s this scale that was developed by NASA, and is now currently used throughout the defence and aerospace industry that describes the maturity of technologies ranging from zero to nine, that kind of describes, you know, all the way from basic research to, you know, real world operation in organisations like DARPA, IARPA -

Fin 41:30

Just to be clear, so advanced research projects in the US are currently government funded for defence applications?

Ajay 41:37

Defence and intelligence, yeah. So DARPA is the ‘Defence Advanced Research Projects Agency’. And then IARPA is the same but for intelligence. And that’s where I think a majority of the truly ambitious projects get their funding. Within corporations, within industry, there are also pretty exciting R&D projects happening. The incentives there are slightly different, in that they’re looking for things that they can take to market pretty quickly. I guess for some of the moonshot-lab type organisations, for example, some of the big tech companies, they have a slightly longer time horizon, but even then it’s on the order of, you know, a few years before they want to try to bring something to market, at least, you know, for some of them. And this ability to kind of take this long time horizon and be very ambitious, and making investments, is something that is, I guess there’s kind of this gap for organisations, you know, outside of academia and industry and government that can accomplish that.

Janvi 42:58

What is the timescale for DARPA, and government organisations for this kind of thing?

Ajay 43:02

Yeah, I guess my understanding is that they take a pretty long term view. So I guess the technologies that they look at depend on components that are fairly recently developed, and I guess I don’t actually know whether there’s like a specific time period that they’re looking at for some of the projects. I think it probably depends on the given project, but looking, you know, decades out at some of the most exciting emerging technologies.

Fin 43:37

Yeah. I wanted to ask a bit about academic institutions. So if I’m at, you know, college, I hit on something pretty interesting, in my PhD, maybe I want to start some kind of spin off that lives in the university. What’s the funding story normally going to look like there?

Ajay 43:58

Yeah. So I guess, if you’re in the US, and you have an exciting technology that could be commercializable, or some other discovery, and you’re starting in academia, there are a couple of types of grants you can apply for from the SBIR programme. The Small Business Administration has both the SBIR and STTR programmes. So SBIR is Small Business Innovation Research and STTR is Small Business Technology Transfer. These allow principal investigators and small teams of researchers to get grants from the US federal government to develop their technology towards the goal of commercialising, often with the aim of providing these technologies to government customers.

Fin 45:02

Okay. I am presuming that these grants have maybe like a shorter timescale attached to them as well.

Ajay 45:11

Yeah. Typically a couple of years.

Fin 45:14

Okay, cool. Got you.

Janvi 45:16

I don’t know if this is right. But it sounds to me like if we’re looking at these three groups where there’s academia, there’s commercial enterprise, and there’s government projects - with academia, and commercial projects, they both have sort of short timescales. But with commercial projects, you want to produce something that is kind of sellable within a couple of years. But with academia, you can do research that’s sort of very basic. And that sort of stacked on top of other grants that are a couple of years, can end up being like quite long bits of time to do like fundamental research, or research that leads to sort of more interesting conclusions over time. So it also seems like a longer term investment. But the mix between these two sounds like that then becomes government grants. Does that sound kind of correct?

Ajay 46:08

Yeah, I think academia certainly allows for this kind of ambitious, longer term thinking. Things that are pre commercial and could potentially be, you know, public goods, very focused on conceptual breakthroughs or open ended exploration. But I think some of the limitations make it such that it’s difficult to take things from academia and actually develop them outside of that context. So there’s not a lot of, I guess, support for the post project transition to commercialization. There’s also, it’s a slightly different working environment than what you might expect from a startup or corporate R&D lab. And that structure comes along with a different set of incentives.

Fin 47:14

Cool. So I guess just to take stock: so we’re interested in, like you mentioned, kind of pre commercial, public goods-y research projects - the kind of project which you need, presumably a bunch of funding over a longish timescale to really get off the ground. And you’re just describing the funding landscape, and it sounds like, okay, there are ways to do this with government grants inside academia, but the working environment is a little different sometimes. Sometimes to transition to something commercializable, like to exit, can be a bit different. Then you have corporate R&D kind of labs, but they operate over often shorter timescales. If you want a longer timescale, ambitious stuff, then maybe the best place is the Advanced Research Projects inside the US, but then maybe you need to be able to tell a story about defence or intelligence applications. And that’s not gonna apply to every useful technology that’s just good for the world. Which leads to this question, right, which is: are there any new, interesting, better ways that these exciting or useful technologies can get funded?

Ajay 48:32

Yeah. So Adam Marblestone at Convergent Research, which is part of Schmidt Futures, has put forth this concept of focused research organisations. And the idea there is that it’s a new kind of structure that - like a startup - has this, you know, close knit, fast moving, mission driven kind of organisation, led by a CEO, but they take on very risky projects that you might expect out of more of a DARPA funded type project. I guess, pretty crucially, as different from a corporate lab or from academia, they are kind of shielded from a lot of the incentives from those structures, and so can focus just on this goal. And it’s sort of this fixed duration, time limited effort, as well, where you say we have this very clear goal that we want to achieve, and we’re going to spend, you know, this duration of time attacking this problem in this very focused manner. Yeah, definitely would encourage you to have Adam on the podcast and talk more about that.

Fin 49:54

That sounds great. I guess one quick question to follow up on that: so you said that focused research organisations shield you from certain incentives; I guess you’re implying that these could be kind of distortionary or bad incentives. But what exactly are they?

Ajay 50:09

Yeah, so I guess in a corporate lab you have shareholder incentives, and that means the things that you develop not only need to get to market fairly soon, but they also need to do well on the market; they need to be things that there’s a significant amount of demand for and also that aren’t too outside of the realm of what the company does. So you can imagine, for example, in an aerospace company that’s in their R&D lab, potentially developing medical devices, you know, that could be really exciting work, they could be making really great progress, but if the shareholders don’t want to see that - they just want, you know, this stable, you know, aerospace company in their investment portfolio, and they see this company doing like weird R&D things - they’re not going to be super happy with that. So there are, yeah, kind of weird incentives that you have to keep in mind there.

Fin 51:11

Okay. This reminds me of a Mitchell and Webb sketch set inside the Garnier Laboratoire where they’re like coming up with perfumes, and then someone discovers a cure for Alzheimer’s, and then Monsieur Garnier walks in and learns this news and like smashes the test tube because it’s not relevant for the Garnier Sleek and Shine range this year or something.

Ajay

Sounds about right!

Worst case scenarios / refuges

Janvi 51:36

So I guess earlier, we spoke a little bit about sequencing things and detecting outbreaks early before they become really bad. But there are some worlds where these outbreaks are so bad that we might not be able to contain them. What is the difference in terms of what we do with an outbreak that we might be able to contain, and an outbreak that we might not be able to contain?

Ajay 51:59

Yeah. So if we detect something early and characterise it and identify that it’s actually something worth worrying about, but for whatever reason we’re not able to contain it - whether that’s a result of institutional capacity for response, or whether it’s some characteristic of the pathogen or something else - the hope is that the extra time we buy with early detection allows us to take other actions that kind of guard against worst case scenarios. And those are both getting people to a position of relative safety and working on developing medical countermeasures to deal with the threat. And I guess, there you can think of getting people to a place where they can work safely in a few different ways. Broadly, there’s non pharmaceutical interventions in general, and then guess maybe a subset of that is physical defences from pathogens. So that can be anything that encapsulates a person and keeps them separate from threats in the external environment. I guess the most obvious form of that would be PPE - you know, something you wear on your face or over your whole body. And then I guess more generally, there’s ways to encapsulate multiple people - these can be buildings that offer similar sorts of protection, but in a context that allows people to sort of function relatively normally. So I guess one way of thinking about this is, if you consider a BSL-4 Lab, which is designed to contain the most lethal pathogens, and prevent them from escaping - if you sort of inverted that and treated the outside world as the high containment area, and the interior spaces the safe area. That is one way of thinking about how to design such a structure. And you could actually imagine that you might want the people who are developing medical countermeasures to be kept in this position of relative safety so that they can work on developing and testing MCMs [medical countermeasures] and have access to stockpiled supplies. You can imagine in perhaps some of the more catastrophic scenarios that supply chains might not be functioning and physical security can’t be guaranteed, and so then how do people actually get access to the resources that they need to develop MCMs? And you may want to be able to provide that for them in such a structure.

Fin

MCMs, to be clear, are medical countermeasures?

Ajay

Medical countermeasures, yeah - vaccines, therapeutics.

Fin 55:20

Got it. Got it. So you want to find, I guess, the most important people and facilities for getting things back on track, and really make sure that they’re effectively isolated against the pandemic in this kind of worst case scenario.

Ajay 55:35

That’s the thinking, yeah.

Fin 55:36

Got it. Okay, so I guess we’re talking about worlds in which, unfortunately, it’s too late to contain or isolate a pandemic in one place. And perhaps it’s like, especially virulent, so it’s fairly worst case. And then you’re describing refuges as a kind of last line of defence or something.

Ajay

Yeah.

Fin

So can you just say more about what these refuges might look like?

Ajay 56:02

Yeah, definitely. So when we say refuges, I think we mean maybe a few different things, and it’s maybe worth disentangling some of those. So the one that I described is maybe like a countermeasures type refuge, where we have scientific personnel who can develop MCMs in a position of relative safety. There are other types that, you know, are hardened against maybe a wider range of threats outside of just biological threats. Maybe they contain information or tools that people can use once they reemerge, but I guess I’ll kind of leave that aside for now and talk just about medical countermeasures type refuges. So what would those look like? You can imagine that for each of the inputs to the structure - so those can be air, water, food, other supplies, and then obviously, the people who are going in - you want to be able to decontaminate and then verify that those inputs are decontaminated.

Fin 57:17

Sorry. Just to be clear, are we talking about something like a large building, for instance?

Ajay 57:20

Yeah. It could just be, you know, a normal building, but it has certain features that ensure that pathogens outside can be kept out. And so that might look like having at the entrance for people, you have airlocks, with, you know, sterilised water showers as you enter, and then once you are inside, you have kind of isolated living quarters that allow you to quarantine, and then to enter the interior of the structure you put on some PPE that is resistant to chemical decontamination. And you can go through an airlock with a chemical shower. And then if you want to go back outside, you can go through the other airlock and go through another chemical shower. And so I guess, yeah, there’s like a whole range of things you can do. Obviously, you want to screen people before they enter and make sure that they’re not infected. And you can do that using some of the sequencing technology we were talking about before, in theory. And in order to get all of the supplies in - obviously you want to have things stockpiled in there, particularly things that allow you to develop MCMs - but also you’ll need to bring things in and deal with kind of a range of possible transmission routes that the might you might want to guard against.

Fin 58:46

Yeah, in terms of the supplies in these refuges for preparing MCMs, what kinds of things are easiest to overlook that just seem really important?

Ajay 59:03

Hmm. That’s a good question. I mean, probably given that these would be environments with limited resources - you wouldn’t have access to, you know, the full suite of pharmaceutical development tools and equipment and people with the relevant expertise - probably the most important thing would just be the capacity to synthesise DNA to develop nucleic acid based therapeutics. Obviously, you know, in an ideal world, you’d have more robust capabilities than those but that’s yeah, probably the bare minimum.

Fin 59:51

Yeah, I guess another question is when I think of ‘refuges’, my mind goes towards post apocalyptic bunkers, right - it’s like lead clad, lined with tins of baked beans. How are those pictures of, you know, bunkers different from what would actually be realistically best?

Ajay 1:00:14

Yeah. I mean, I guess it depends on what what your threat model is, I think some of those kind of more intense looking bunkers are hardened against things that look more like, you know, nuclear weapons and obviously, many of those if not all of them would not be hardened against like a direct strike from a nuclear warhead, but they’re, I guess, designed for a different type of threat. And maybe the thing that we want is not, you know, just one structure that is hardened against every possible threat, but maybe things that are purpose built for different things that we’re worried about.

Janvi 1:01:01

I’m just gonna say there seems like there’s a range of intensities here, where you could imagine that some of these things that you’ve proposed could exist kind of passively, sort of within the homes we already live in. So sort of like ventilation and having strengthened PPE around, in the case where sort of any bad outbreak starts to happen. And then there are the sort of things that Fin mentioned - this more intensive type bunker. And there seems to be a trade off in terms of how bad the thing is, and how intense the intervention or bunker seems. Is that true?

Ajay 1:01:43

I think that’s probably right. Yeah. There certainly are built environment improvements that one could make in just normal buildings, like normal office buildings, schools - schools would probably be one of the best places to do that, given how high transmission rates are in schools - to just retrofit HVAC [heating, ventilation, and air conditioning] systems with HEPA filters or upper air UV disinfection. Yeah, there’s a range of lower cost retrofits that you can do for existing structures. And those definitely seem worthwhile, but I think for some of the more extreme threats and extreme scenarios, then something more robust is probably needed.

Retrofitting and other promising countermeasures

Fin 1:02:27

Yeah. I mean, tell me if I’m wrong, but it does seem like there are a bunch of low hanging fruit with just the retrofitting of existing buildings, right. So a HEPA filter is just a block of sponge with a fan grabbing air through it, right. So presumably, it’s not super expensive to just install that permanently in HVAC systems, ventilation systems. And then you mentioned something to do with UV that I hadn’t heard of - what’s the idea there as well?

Ajay 1:02:56

Yeah. So the idea with UV is that irradiation can damage and destroy a lot of pathogens. And you can accomplish that with UV kind of on the more intense end of the spectrum, and you can also do that with gamma radiation. But yeah, no, I think you’re totally right. I think that there’s a lot of low hanging fruit in retrofitting existing facilities, there have been a few different organisations that have published guides to retrofitting buildings to protect against biological threats. So there’s ‘system recommissioning’, where you can do air flow testing and balancing, calibrating sensors, testing dampers, etc., tightening the envelope of the building to eliminate leaks, pressure driven airflow control, that sort of thing. One of the larger expenses with retrofitting HVAC systems, for example, is that if you put a really good filter in the system, it requires a lot more power to drive air through the filter. So there’s kind of an added expense there. And as mentioned, yeah, there’s also a trade off, you know, if you want to reduce your power consumption cost, then maybe using a filter that has a lower percentage of particulates -

Fin

Yeah there’s some trade-off there.

Ajay

Yeah, exactly.

Fin 1:04:28

Yeah, that makes sense. Maybe this is too convenient, but it also just seems really good to have cleaner air in general. So it seems like there are some effects on cognition from carbon particulates and from CO2, and it could just be a win-win if we finally get serious about making sure our air is clean in the same way that we just assume that all our water is pure and clean. Yeah, there’s just like multiple reasons that could be good.

Ajay 1:04:57

Absolutely. I mean, if you think about the way buildings are constructed, everything is built with fire suppression in mind. There are a lot of expenses that go into constructing buildings that ensure that, you know, a fire in one place can’t just burn down the whole structure - the furniture we use has fire retardant materials inside of it, there are all these expenses that we build in assuming that fire suppression is something that is an absolute necessity. We have no such equivalent for pathogen transmission suppression. And I think we need to come up with standards for our built environment, that take that into account, that ensure that our air is clean, and that there are, you know, tools that ensure that people can be kept healthy inside of a building.

Fin 1:05:56

This isn’t me doing serious history, but you can imagine some gloss on the way things happened where before, you know, something like germ theory of disease and, you know, appreciating the importance of purifying water, it kind of would have seemed maybe a bit OTT to care so much about, for instance, having standards for like, pure clean water, and also all the expenses involved in setting up proper sanitation systems and sewage systems. And clearly, nowadays, it’s pretty obvious that this stuff matters. And presumably, you could anticipate some similar shift in expectations as to what is, you know, bare minimum kind of important with respect to air as well. It seems like maybe we’re kind of at the inflection point of that kind of transition now that it’s actually feasible to do stuff about it.

Ajay 1:06:50

I think that’s right, yeah. I would definitely be excited to see, you know, a lot more work on figuring out, I guess what standards we need to set both in terms of the quality of the air, and then the tools that we use to achieve that. And then once we figure that out, proliferating those standards, and making it such that, you know, all new buildings need to keep that in mind.

General disasters preparedness

Fin 1:07:10

Cool. So we’ve been talking about, roughly speaking, disaster preparedness on a kind of institutional level - so how do we make sure our buildings are resilient? How do we come up with plans in advance to use them? But you can also do things to prepare on a personal level, right? At least a few things. So I was wondering if you could speak to any of the kinds of quick or relatively inexpensive things that individuals can do to just get prepared for, you know, low likelihood, but potential worst case scenarios? And one obvious example, unfortunately, is that at the time of speaking, you know, it might pay, especially if you live in a big city, to just take like a moment to consider how to prepare for a nuclear exchange in particular. Not because it looks, you know, extremely likely, but just because the possibility is currently fairly, kind of all too salient. So, yeah, I wondered if you could speak to disaster preparedness in general, and then to that specific example, as well?

Ajay 1:08:18

Yeah, definitely. I think, you know, I guess, for better or for worse, over the course of the pandemic, we’ve seen personal disaster preparedness come more into the mainstream. And there are a lot of common sense things that you can do to prepare for disasters in general. And I think those things apply to probably, you know, like 98%, very roughly speaking, of the scenarios you might want to guard against. And I’ll note that, you know, I’m certainly not an expert in nuclear exchanges, whatsoever, but I think there are probably some things that you can do there as well, that are pretty similar, and then some things you might want to do there specifically. I guess at a high level, you want to make sure that you’ve got the basics taken care of - that you have sufficient food and water supplies at home to be insulated from any shocks to those supplies. Make sure you have, you know, medical supplies so that if needed, you can self administer medical care. There’s, you know, kind of a range of things beyond that you can think about, but I think those are kind of the most important ones. And then there’s also this question of, you know, in the event of a disaster, do you stay at home or do you go elsewhere? And usually, I think the right answer is to stay at home unless there’s some strong reason that you need to leave. It’s pretty disruptive to uproot yourself and go somewhere else. It makes it harder to make sure you have all the supplies that you need, you have the kind of support network you need. But in some scenarios, you know, it could be worthwhile, provided that you’re able to ensure you have access to all those other resources that you need. In general, for disaster preparedness, I would point people to - there’s a really good website called theprepared.com. It’s just very rational common sense disaster preparedness advice. They do a pretty good job of that. There’s also - my context is in the US - FEMA has some resources online on how to prepare for different types of disasters. ready.gov is a good place to go. And then for nuclear exchange preparedness, in particular - a little more complicated there. I did recently see a really good EA forum post by another Fin - Finan Adamson, it looks like.

Fin

Rivals.

Ajay

Yeah. The anti-Fin. The title of this post is just ‘Nuclear Preparedness Guide’. And this is a very well researched, thorough post. And I can see that a lot of information has been drawn from some of these guides, from FEMA and otherwise. So I certainly would recommend taking a look at that.

Fin 1:11:26

Super. And we’ll link to all those things in the write up.

Ajay’s career

Janvi 1:11:29

Yeah. Okay. So it sounds like there’s a range of really cool sort of biosecurity interventions you’re working on. But one question I had is: I’m kind of curious as to how you became interested in all this stuff and what you were working on before, and what that did to sort of lead to this?

Ajay 1:11:46

Yeah, definitely. So my background is in electrical engineering, I’d spent some time doing medical device research in neural microsystems for a little while, and then afterwards, non invasive human computer interfaces, and spent a few years at a startup that I co-founded with some good friends, making wearable devices, with some customers in the US federal government and in the private sector. And then most recently, I spent a few years actually at Amazon, both on the AWS [Amazon Web Services] side, working on their machine learning platform, and then also within their moonshot lab, which is called Grand Challenge, working on a project in healthcare and life sciences. And as far as how I got into the biosecurity work specifically - it was really at the start of the pandemic that I was, you know, sitting there, in my apartment, in quarantine, trying to figure out what the hell happened? How was it possible that a new pathogen can emerge and we don’t have an early warning, we don’t have any capability to guard against it. I just kind of assumed, in my naivety, that there were equivalent systems to what we have for hurricane early warnings or other types of natural disasters. And when I started doing some digging and reading about it online, I very quickly realised that no such system exists. And I just thought that was crazy! And at first brush- I guess I sort of started from the thinking, like, let’s assume that there’s a high consequence pathogen that emerges in humans, what do we need to do about it? Like, how can we stop it as quickly as possible? And that led me to looking into these pathogen agnostic diagnostics, which then led me to metagenomic sequencing and I spent some time trying to figure out, you know, can we just build something right now that can be deployed? And after a little while, realised that that’s not possible and spent some time talking to folks that 80K put me in touch with, and others and that’s, yeah, eventually what led me to the work I’m doing now.

Fin 1:14:22

Cool. So here’s one question I had: it sounds like your background - as in your academic background - is primarily not in anything related to biology, or at least not centrally related. So it sounds like it’s mostly an engineering background. It also sounds like you’ve pretty successfully switched from various kinds of engineering flavoured projects into working on doing research in biosecurity. So I’m curious how easy that was to do. How easy was that switch to make?

Ajay 1:14:57

Yeah, um, I guess it’s sort of hard to say. I’m not sure what someone might be able to sort of replicate in the steps that I took. But I will say that in general, biosecurity does need many more people with engineering backgrounds. And I think that’s probably why it was possible for me to do that. There are a lot of things that need to be built, to kind of develop defences against biological threats. As mentioned, there’s this concept of an early warning system for emerging pathogens, there’s also these physical defences against biological threats - whether that’s PPE or other improvements to the built environment, other you know, kind of structures you might want to engineer. So there’s, I guess, a pretty wide range of things that someone with an engineering background can slot into. And I think the range of possible projects for engineers is starting to become more clear. So yeah, I would certainly encourage anyone out there who’s listening who has a background in engineering, whether that’s medical device engineering, mechanical engineering, civil engineering, you know, physics, material science, anything kind of in that flavour of discipline, certainly should reach out if you’re interested in working on biosecurity, because we need many more people like you.

Recommendations

Janvi 1:16:35

Are there any resources in particular that you would recommend to these sorts of prospective biosecurity engineers to read or look into to help them with their transition?

Ajay 1:16:46

Yeah, definitely. There are a few EA forum posts that come to mind. One from Will Bradshaw kind of describing the range of engineering expertise that’s needed - and all these will be linked in the show notes - and there’s another one from Andrew Snyder-Beattie and Ethan Alley kind of describing these concrete biosecurity projects that can be done. And I think both Tessa Alexanian and Chris Bakerlee have some good reading lists on the EA forum for getting up to speed on biosecurity in general. So I would certainly encourage people to take a look at all those. And then of course, if you’re not familiar with the problems and biosecurity overall, then there’s also the 80K guide on global catastrophic biological risks. That’s very much worth a read.

Fin 1:17:41

Super. And as mentioned, we’ll link to all of those things in the write up. All right, let’s do some final questions, then. One question we ask everyone is: is there any kind of research or other work you’d be especially excited to see people get involved with, and this is bearing in mind that maybe some people listening to this might actually have an opportunity to help out on it?

Ajay 1:18:07

Yeah, that’s a great question. I guess, as previously mentioned, folks with engineering backgrounds, certainly should feel free to reach out and jump into some of these projects, you know, maybe developing better PPE, better methods for decontamination - kind of this whole range of projects related to things that need to be built. And people who don’t have engineering backgrounds certainly should take a look at these reading lists as well because there’s a lot of work to do there as well. And, you know, once you’ve kind of read all the material, there are people who will be happy to help you, happy for you to get in touch and get involved.

Fin 1:18:53

Okay, awesome. Let’s do final final questions. Here’s one we ask everyone which is: what three books or films, podcasts, whatever would you recommend to someone who is listening to this and really wants to get stuck in to learn more?

Ajay 1:19:08

Yeah, so I guess maybe I’ll grab a little bit from Tesla’s reading list. The Apollo Program for Biodefense Report is a really good one that Jacob Swett spent a lot of time putting together along with some colleagues and is certainly a good place to start. There are some really fascinating books on biosecurity generally that are worth a read. There’s The Dead Hand, which is a book about the Cold War generally, but there’s kind of a focus on Biopreparat which is the Soviet biological weapons programme. Couple more books in that general vein - there’s Biosecurity Dilemmas is also a good one. There’s also Deadliest Enemy - Michael Osterholm’s book is a great one. Yeah, definitely check out some of the reading lists and see what jumps out at you as interesting.

Fin 1:20:14

Super, and again, we’ll link to all those things. Last question is: where can people find you and anything you’re working on online?

Ajay 1:20:21

Yeah. I am on Twitter. I haven’t posted in, you know, maybe a few years, but at some point I’ll change that and start posting things instead of just spamming my friends with messages on signal. And my Twitter handle is just my name - @ajaykarpur. Yeah, outside of that, people can email me. It’s hello@ajaykarpur.com. .

Fin 1:20:49

All right, Ajay Karpur - thank you very much!

Ajay

Thanks for having me!

Fin

And Janvi - thanks for joining us.

Janvi 1:20:53

Yeah, of course. Thanks so much!

Outro

Fin 1:20:55

That was Ajay Karpur on metagenomic sequencing. As always, if you want to learn more, you can read the write up at hearthisidea.com/episodes/karpur. There you will find links to all the books and resources that Ajay mentioned, plus a full transcript of the conversation. If you find this podcast valuable in some way, one of the most effective ways to help is just to leave a review wherever you’re listening to this - Apple podcasts, Spotify, wherever. You can also follow us on Twitter - we are @hearthisidea. If you have constructive feedback, there is a link on the website to an anonymous feedback form. There’s also a star rating form on the top and the bottom of each write up. And you can send in suggestions, questions and whatever else to feedback@hearthisidea.com. As always a big thanks to our producer Jason for editing these episodes, and also to Claudia for writing full transcripts. And thank you very much for listening!