#104 – Dr Pardis Sabeti on the Sentinel system for detecting and stopping pandemics

Taken from an animation that depicts how Cas13 — a CRISPR-associated protein — may be adapted to detect human disease, via a diagnostic tool called SHERLOCK.

When the first person with COVID-19 went to see a doctor in Wuhan, nobody could tell that it wasn’t a familiar disease like the flu — that we were dealing with something new.

How much death and destruction could we have avoided if we’d had a hero who could? That’s what the last Assistant Secretary of Defense Andy Weber asked on the show back in March.

Today’s guest Pardis Sabeti is a professor at Harvard, fought Ebola on the ground in Africa during the 2014 outbreak, runs her own lab, co-founded a company that produces next-level testing, and is even the lead singer of a rock band. If anyone is going to be that hero in the next pandemic — it just might be her.

She is a co-author of the SENTINEL proposal, a practical system for detecting new diseases quickly, using an escalating series of three novel diagnostic techniques.

The first method, called SHERLOCK, uses CRISPR gene editing to detect familiar viruses in a simple, inexpensive filter paper test, using non-invasive samples.

Rapid diagnostic tests [are a] terrific technology, but usually it takes about six months to develop a new one because the proteins are a little more bespoke… Whereas the genome sequence, it’s just literally like a code, you just put it in and you immediately can target… You type it out and you have it going.

If SHERLOCK draws a blank, we escalate to the second step, CARMEN, an advanced version of SHERLOCK that uses microfluidics and CRISPR to simultaneously detect hundreds of viruses and viral strains. More expensive, but far more comprehensive.

Most infections all look the same — Lassa looks like Ebola, which looks like malaria, which looks like typhoid, and other things at varying stages. So you don’t want to have to know exactly what you’re looking for in a lot of cases; you want to do a broad differential that you test for.

If neither SHERLOCK nor CARMEN detects a known pathogen, it’s time to pull out the big gun: metagenomic sequencing. More expensive still, but sequencing all the DNA in a patient sample lets you identify and track every virus — known and unknown — in a sample.

Those are the kinds of technologies that we can have in the kinds of labs that we could have in every country on the planet, and even in a lot of regional centers. Then if something comes up and all the standard tests that you’ve run don’t know what it is, you can basically try to put it through.

If Pardis and her team succeeds, our future pandemic potential patient zero may:

  1. Go to the hospital with flu-like symptoms, and immediately be tested using SHERLOCK — which will come back negative
  2. Take the CARMEN test for a much broader range of illnesses — which will also come back negative
  3. Their sample will be sent for metagenomic sequencing, which will reveal that they’re carrying a new virus we’ll have to contend with
  4. At all levels, information will be recorded in a cloud-based data system that shares data in real time; the hospital will be alerted and told to quarantine the patient
  5. The world will be able to react weeks — or even months — faster, potentially saving millions of lives

It’s a wonderful vision, and one humanity is ready to test out. But there are all sorts of practical questions, such as:

  • How do you scale these technologies, including to remote and rural areas?
  • Will doctors everywhere be able to operate them?
  • Who will pay for it?
  • How do you maintain the public’s trust and protect against misuse of sequencing data?
  • How do you avoid drowning in the data the system produces?

In this conversation Pardis and Rob address all those questions, as well as:

  • Pardis’ history with trying to control emerging contagious diseases
  • The potential of mRNA vaccines
  • Other emerging technologies
  • How to best educate people about pandemics
  • The pros and cons of gain-of-function research
  • Turning mistakes into exercises you can learn from
  • Overcoming enormous life challenges
  • Why it’s so important to work with people you can laugh with
  • And much more

Get this episode by subscribing to our podcast on the world’s most pressing problems and how to solve them: type 80,000 Hours into your podcasting app. Or read the transcript below.

Producer: Keiran Harris
Audio mastering: Ben Cordell
Transcriptions: Sofia Davis-Fogel

Highlights

SENTINEL

Pardis Sabeti: We call the three tenets detect, connect, and empower. And there’s also, of course, the piece of overcoming, which is building those countermeasures. But what we were focusing SENTINEL around is just what we as a community can do. In essence, it’s building technologies to detect viruses in lots of different settings. So having these sequencing technologies that can detect and characterize a novel virus with this sort of advanced technology, on ground, in every country, hopefully in every major setting. If you see a new infection, you don’t know what it is, to be able to read out and try to uncover the origin. But then you need to be able to convert those to molecular tests that can deal with a broad differential.

Pardis Sabeti: Most people come in, they don’t know if they have COVID, or the flu, or RSV, or a host of other things. So we need to have other kinds of diagnostics that can test for a number of different viruses simultaneously in individuals that are ill, and then when there’s certain things that we know are high probability, we can turn those into point-of-care, even point-of-need, like right in the clinic, in the home, if possible. Tests that could just really pick it up closest to ground zero, right? Because you really don’t want people coming into some central place when they may have a deadly virus, because likely either they will get it or give it while they’re there.

Pardis Sabeti: So it’s to be able to have ways of detecting and diagnosing these threats wherever, and then to be able to connect that information in real time. We don’t use big data enough in medicine, often it’s just a relationship between you and the doctor and what the doctor’s recall is about what they’ve seen. Whereas you really should be able to put the symptoms into a system and pattern match, and then have the doctor overlay their knowledge on that. We don’t use that big data enough in medicine generally, but particularly in infectious disease where there’s another overlay — the network, who you are in contact with really matters — and so the other thing is making sure we can integrate the data across that piece. If there’s a lot of cases of COVID coming up in the area that you were in, there’s a higher probability the person you’re seeing has COVID. And so how do we use that information in real time and connect it?

Pardis Sabeti: And then the last piece is empower, and empower is really about empowering every actor in the system. A lot of our work has been around frontline workers, making sure they have the tools and technologies. We’re basically giving them tools so that they get real-time actionable data that also incentivizes them to give the best data. The better data in, the better information out. And so we’re trying to build these tools that work anywhere, but also through education, we’ve developed a number of programs, we’ve trained about 1,000 frontline workers so far in genomics, bioinformatics, diagnostics, other kinds of things like project management, all the things that you need to be able to run these programs, we’ve been working on building out that capacity. So that’s sort of the tenets of detect, connect, empower, and all of that is to basically get a global community working together, including every citizen, to really stave off and keep eyes on the virus and keep connected and positive with each other while we come up with these vaccines and therapies and other ways of stopping it.

SHERLOCK

Pardis Sabeti: The way CRISPR works is that it has an exquisite system of just detecting a particular sequence, and then performing a very precise cut. And we’ve been able to use that to great effect in a lot of molecular medicine. But it’s interesting from a diagnostic standpoint, there’s particular enzymes, Cas13 and more recently Cas12, that were discovered to have something called collateral effect, which means that the cut that happens has to identify a very precise sequence, so that’s where you get that sort of specificity of the cut. But then once it sees that thing, it starts cutting everything. It says, “Okay, it’s go time, we got to start cutting.”

Pardis Sabeti: And basically, Feng Zhang, who is one of the other pioneers of CRISPR is a colleague of mine, and he was working with Jim Collins, trying to use this as a diagnostic with a very a cool effect. Basically, seeing that if all that cutting is happening and these enzymes are doing that, you could then create this thing where you tie this whole system to a fluorescent readout and a quencher. So, once the thing starts cutting wildly, it’ll cut these fluorescents from the quencher and it’ll create a signal. And there’s a lot of different readouts you can use for this, including ones that can be colorimetric, essentially they can be read out on paper. So, that’s very, very powerful for getting a field deployable test. So, what’s great about the CRISPR system, in that context, and again, it’s one of the multitude of really new, exciting technologies, but why we decided to invest in that space and we like it is: It’s a molecular test. Meaning that you can basically just know a sequence and put in a sequence and make a new diagnostic immediately, that can be read out on paper.

Pardis Sabeti: So, there are these things called rapid diagnostic tests, these point-of-care tests, that have been available, but most of them are based on the protein of a virus, like the Abbott Binax where it’s an antigen capture test, it will pick the protein of the virus. Terrific technology, but usually it takes about six months to develop a new one because the proteins are a little more bespoke. They have these certain properties, you have to figure out exactly which will work and how. Whereas the genome sequence, it’s just literally like a code, you just put it in and you immediately can target, you type it out and you have it going. So, I think from that standpoint, it’s really powerful because of the fact… the PCR works in that way that it’s fast to deploy, the second the virus had it, people had PCR tests going, but this is something that’s fast to deploy, even arguably a little faster to develop than PCR, because it’s conditions are very uniform, but it can be rolled out and taken to the field and also taken to this massive multiplex that you mentioned.


Rob Wiblin: So, just to paint a picture of how useful this could be, for listeners, imagine you’re in Nigeria, in a lab out in a fairly rural area, someone comes in and it looks like they have, hypothetically, it looks like they have malaria, and you’ve got this pretty straightforward test, you’ve got the little SHERLOCK… I’m not sure what you call it, the little plastic thing, you’re going to take a blood sample, do what I think is a relatively straightforward PCR process, there’s a way of amplifying the genetic material of the pathogen in these rural areas, and then I think within hours, maybe within half an hour, you can actually say whether the person has malaria or not, which is a big step forward on having to take a sample, send it away, transport it somewhere else, and probably it’s quite a lot more expensive to do it using the normal PCR test.

Pardis Sabeti: Yeah. That is the aspiration.

CARMEN

Pardis Sabeti: CARMEN is that pairing with microfluidics, so it’s a combinatorial process. CARMEN is after Carmen Sandiego — still staying with the detective theme — but essentially it’s pairing the power of CRISPR technology with these microfluidic and miniaturization technologies to be able to do this at high scale. Most infections all look the same — Lassa looks like Ebola, which looks like malaria, which looks like typhoid, and other things at varying stages. So, you don’t want to have to know exactly what you’re looking for in a lot of cases; you want to do a broad differential that you test for. This is about having technologies that you could have anywhere in the world that you could use to test for a number of different viruses.

Pardis Sabeti: There are other groups that have developed methods like this. I’m pretty open to wherever the technology comes from. We just look to see where we can make a contribution; other groups may have technologies that end up ultimately working better. What we’re excited about with what we’re doing right now, and the need that we see, is that we haven’t seen things that can do lots of viruses on lots of samples, to be able to scale that. There are technologies that may do that, may do a differential panel, but usually still like a sample at a time or something like that. So, we were really excited that you could say, “Okay, we could take a hundred samples and run them.” I think right now in the instantiation that we just submitted to the FDA it was something that could run about one machine, about a thousand samples a day for a panel of viruses. So, that’s exciting to just be able to do that at scale.

Metagenomic sequencing

Pardis Sabeti: So metagenomic sequencing is just where the ability to read out the sequence of anything in the sample is very powerful, where in COVID amplicon-based sequencing has been really potent. It’s a little bit more directed, you have to actually have guides for the thing you’re looking for, but you can put guides for a lot of different things. So there’s just varying ways of getting more and more of a broader read of what is in the sample. And those are the kinds of technologies that we can have in the kinds of labs that we could have in every country on the planet, and even in a lot of regional centers. Then if something comes up and all the standard tests that you’ve run don’t know what it is, you can basically try to put it through.

Pardis Sabeti: It’s not always possible. The virus has to be in the sample that you’re looking for, and microbes in general are pretty stealthy. Something like Zika is causing a lot of damage, but is only there in a very small amount and for only a short period of time. And so they’re what they call “hit-and-run” microbes: they come in, do their damage, and then they disappear. So it should be noted that there’s still a lot of things that — even with all of this technology — might be missed. But that’s why we also use lots of different kinds of… There are technologies like serology-based technologies that can be better at finding the hit-and-run microbes.

Pardis Sabeti: But even when the microbe comes in, does its damage, then disappears, your immune system’s still reacting to it. And we can read out what your immune system is doing to figure out what the perpetrator was. But again, it’s all being a disease detective. And all these different clues… The way we’ve built SENTINEL, it’s not a place to test our CRISPR-based technologies. It’s a place to test and deploy and utilize the best technologies out there. And we do believe that it’s a whole series of technologies. So I think the innovation we’re trying to put forward is as much about how we do these things and how we create these hubs, where many people can come and bring their technologies and we can be honest about what’s working. We want to be able to give people feedback and say, “Hey, we tested your technology. It’s good for this. It’s not good for this. These are what you could do to advance it. This is how we’re going to use it.”

Pardis Sabeti: So trying to do the most unbiased process we can. Ultimately what motivates me when I get up in the morning is solving the problem. It’s not about the technologies I’m advancing to solve the problem, but solving the problem and being part of that process.

Failure modes for this technology

Pardis Sabeti: We’ve thought about this in varying ways. There are many different failure modes for this technology. Misuse will be a really important thing to watch for, because a lot of the technologies… We haven’t talked about the fact that things like mobile applications and Bluetooth and geolocation, all of those things are also really important and will be beneficial to a system, but have so many potential misuses. I think probably the single biggest challenge to the use of this is protecting against misuse. And at the end of the day, viruses are insidious deadly threats that weaponize your neighbor against you, and if we don’t manage that well, people can get pretty hysterical. And we’ve seen it with HIV, with COVID, with Ebola, so many different cases in which the culture becomes very toxic.

Pardis Sabeti: And so that’s one of the things that we try to build into everything we do. What’s important is this idea of making sure that the systems we put in place are ones that are thought through with regard to how they work for the communities. And that every potential misuse is considered. If you want to really stop a pandemic, you need incredible amounts of visibility. And it’s really challenging when you have all of the next generation of kids out there on TikTok and Instagram and all these places sharing all of this really personal private information with no problem, but then when it’s your location as it relates to potentially spreading COVID amongst the community, nobody wants to talk about it. There’s part of me that’s like, what’s going on here?

Pardis Sabeti: You’re literally giving away the most personal private information, but yet the thing that could actually save a life, you’re afraid to share. But at the same time, there’s a reason for that. It can be so stigmatized. And so what we’ve been doing is we’ve been piloting these types of technologies in very specific settings. When we started developing it, we were developing it for Harvard at Harvard. And the way I pitched it is the Facebook app for outbreaks, that starts within a close-knit community. A place where you can test and see how things go and get buy-in, local buy-in, and show utility to people. But Facebook can be co-opted in all sorts of bad ways, so it has to be protected and nurtured at every step.

Pardis Sabeti: And so I do think that it is about showing the use case in some environments in which there is trust, and then being able to then roll those out to other places while maintaining that level of personal-ness. I often say public health needs to be local. You can’t have somebody in Washington trying to figure out what went down in some school somewhere else. The local janitor will have more information about things, like, “Oh, well actually there’s this bathroom that everybody uses.” You actually want to empower every single person on site to be able to do this in the right way with people they trust, and in a way that builds trust.

Pardis Sabeti: There are many, many challenges, but that’s probably the one that’s foremost in my mind. How do you build these technologies? And how do you roll them out in a way that they are used well and where we remember that the virus is the threat, not each other? It is tough. And I think there’s a lot of work we have to do to build trust in society broadly in order to get there.

Related episodes

About the show

The 80,000 Hours Podcast features unusually in-depth conversations about the world's most pressing problems and how you can use your career to solve them. We invite guests pursuing a wide range of career paths — from academics and activists to entrepreneurs and policymakers — to analyse the case for and against working on different issues and which approaches are best for solving them.

The 80,000 Hours Podcast is produced and edited by Keiran Harris. Get in touch with feedback or guest suggestions by emailing [email protected].

What should I listen to first?

We've carefully selected 10 episodes we think it could make sense to listen to first, on a separate podcast feed:

Check out 'Effective Altruism: An Introduction'

Subscribe here, or anywhere you get podcasts:

If you're new, see the podcast homepage for ideas on where to start, or browse our full episode archive.