Monday, July 6, 2020

Out of the Crisis #13: Alpha Lee on remembering the 2003 SARS epidemic and his opensource COVID-19 Moonshot

As a child, Alpha Lee lived through the SARS epidemic in Hong Kong. The human toll a virus could inflict and the immense economic cost of the emergency response stayed with him. Unfortunately, the rest of the world soon forgot about it. In the years that followed, Alpha dedicated his life to science, pursing degrees in chemistry and mathematics, and starting a research lab the University of Cambridge. His lab specializes in accelerating the discovery of new drug compounds and treatments. Then he started a company called PostEra, which uses machine learning to identify promising compounds for new drug development.

In January of this year, he arrived in the US with his co-founders as part of the winter 2020 YCombinator batch. When the virus began sweeping though the world, they realized the data they had on drug discovery and the building blocks of therapeutics could be applied to finding therapeutics for the coronavirus. That was when they launched their COVID-19 Moonshot project, an opensource invitation to thousands of collaborators from around the world to contribute potential solutions. As Alpha explained, it's "not in any way tied to PostEra's commercial operations. It's solely used to make and test compounds against COVID."

They've collected and sorted thousands of possible compounds, and now have multiple drugs in development in a lab in the UK. Each of them has the potential to help not just now, but in the future. "If the world only looks for short term solutions," Alpha told me, "then you will always get pandemics over and over again."

You can listen to our conversation on Apple, Google, or wherever you like to download podcasts.


In addition, a complete transcript of our conversation follows the show resources below.

Highlights from the show:

  • Alpha introduces himself, gives his background (2:32)
  • Alpha's favorite quarantine tip (3:40)
  • The moment he realized the magnitude of the pandemic and that it would change his life, and how social media was involved in that (4:19)
  • Why he dedicated his life to science and his background and scientific training (7:17)
  • His transition from scientist to founder of a tech startup (10:06)
  • What it was like to be part of YCombinator when the news of the virus broke (11:49)
  • How drug discovery works (12:29)
  • What it takes to build a drug in the real world (15:21)
  • What the conventional discovery process looked like before machine learning (20:00)
  • How doing drug discovery this way can affect the timeline for finding vaccine (21:18)
  • The benefits of a faster iteration cycle (23:39)
  • How open-source accelerates the project and gathers the best chemists (25:32)
  • Description of PostEra and where it was headed pre-pandemic (27:02)
  • What happened in the days after Alpha realized PostEra's work could be useful for COVID (28:27)
  • Why the project was open-source from the very beginning (30:27)
  • The response the the call for scientists to participate (32:29)
  • The genuinely new ideas for therapeutics that wouldn't have come to light with the moonshot (34:55)
  • Explanation of trials to stop the virus's power to replicate (36:36)
  • The timeline and what's needed to move ahead with a viable drug candidate (37:57)
  • Funding and the project's GoFundMe page (39:09)
  • Why drug discovery work is crucial for now and going forward (40:45)
  • Why short-term solutions to a long-term, ongoing problem don't work (42:33)
  • On the need for government funding and science-driven policy (44:24)
  • The need for changes in the private sector side of the equation (47:57)
  • Details on the project's GoFundMe  (50:15)
  • On trying to save the world while living in an Airbnb (51:28)
  • On not learning from the SARS pandemic and what's keeping Alpha optimistic (52:13)

Show-related resources:

Transcript for Out of the Crisis #13, Alpha Lee on a COVID-19 Moonshot 


Eric Ries: This is Out of the Crisis. I am Eric Ries.

The institutions that governed our lives before are under extreme stress, and the cracks and gaps in the system are becoming rifts and fault lines. We have an opportunity in front of us to rebuild and make our world more resilient, more equitable, and to put human talent to better use. This opportunity is so large, we can't afford to waste it.

Alpha Lee is a scientist who has lived through this type of epidemic before. As a child living in Hong Kong through SARS, he saw first hand the human toll a virus could inflict, and the immense economic cost of the emergency response. Unfortunately, like many episodes in history, once the SARS emergency was resolved, this sacrifice and experience was largely ignored, at least by the rest of us. In the years that followed, Alpha dedicated his life to science. He pursued degrees in chemistry and mathematics, and started a research lab at one of the most prestigious institutions in the world. His lab specializes in accelerating the discovery of new compounds and treatments. Last year, he started a company called PostEra, which uses machine learning to identify promising compounds for new drug development.

When Alpha saw the pandemic rolling across the world, he realized he had a chance to fix one of the problems left unsolved from SARS. When the shelter in place orders began, Alpha found himself living in a tiny Airbnb with his other co-founders, Having just arrived from the UK to do the start up program, YCombinator. He realized that the data he had on drug discovery and the building blocks of therapeutics could be the key to ending the lockdown soon. They took an open source approach, inviting thousands of collaborators from around the world to contribute potential solutions. Their non-profit effort is called the COVID moonshot. They have collected and sorted thousands of possible compounds, and now have multiple drugs that could treat COVID being developed in a lab in the UK. None of this would have been possible without the open source contributions of thousands of scientists around the world. His main message throughout all of this, Let's not waste the opportunity this time.Here's my conversation with Alpha Lee.

Alpha Lee: I'm Alpha Lee, a co-founder of PostEra. PostEra works on medicinal chemistry as a service, powered by machine learning. We are working on a open-source drug discovery project against COVID-19. Before PostEra, I am a group leader at the University of Cambridge, where I lead a group of 11 post-op scholars and graduate students. Before Cambridge, I was a post-doc at Harvard, and in my PhD at Oxford University.

Eric Ries: So, I want to get into your work as a scientist, and of course, the efforts that you're pioneering to help us get out of this mess. But first of all, these are really difficult times for a lot of people. How are you doing? How's your family? What's your quarantine set up like? How's it been?

Alpha Lee: Yeah, it's been good so far. I'm still safe and healthy, which I think is a blessing in these challenging times.

Eric Ries: Amen to that.

Alpha Lee: Yeah. We are in Santa Clara, California, still holding up well.

Eric Ries: Is there anything that has surprised you about being in quarantine or during the lockdown, or do you have a favorite tip?

Alpha Lee: Yeah. My favorite tip is just focusing on the positive news. I think A lot of news is rightly very tragic, but I think there is always a positive component of human courage looking at all the medical professionals sacrificing their lives in the front line. Those are the positive aspects of humanity that I force myself to focus on.

Eric Ries: We'll put a link to Good News Daily in the show notes for those who want to have a moment a day to get out of that negative headspace. Was there a moment for you when you realized this pandemic was going to be a life changing event?

Alpha Lee: I was initially hoping that the pandemic would be a slightly worse version of the common cold or flu. This hypothesis might be the case until we saw the crisis in Italy. First in Wuhan, China, and then the fact that it can spread so quickly in Italy and continental Europe really surprises me and I think a lot of others as well in the scientific community. Then I realized, "Oh no, this is really, really, real."

Eric Ries: Was there a specific moment or day when you realized that this was going to change the course of your life?

Alpha Lee: Yep. Initially when the crisis hit, I thought I was in some passive mode, obviously reading a lot of papers, but not doing much. PostEra as a company offers medicinal chemistry as a service. We got into YC and we thought we’d do drug discovery, never thought that the two would be related, somewhat strangely. I think the aha moment came when we saw a Tweet from a group of scientists working at Oxford, UK, who have basically replicated a piece of very nice work by Chinese scientists, that determined the crystal structure of an important machinery that underpins how the virus replicates. They also did some work showing that you can have very small pieces of molecules that can stop this machinery.

Once we saw this evidence, we thought, "Wait a sec, we can actually do something about it." We have the machine learning technology to take these disparate, small pieces of molecules, and stitch them together to form an actual anti-COVID drug." Then we Tweeted back at the group, and whole collaboration, the whole COVID moonshot started in a Twitter exchange, which then spiraled into an international movement.

Eric Ries: That's a remarkable use of social media.

Alpha Lee: Yeah.

Eric Ries: I got to tell you, I have seen so many examples of that kind of thing, where these technologies that we took for granted before or even derided as not especially socially useful, have enabled people to collaborate in unprecedented ways and unprecedented speed. I certainly... can you imagine when you first heard about Twitter? We used to use it to text in what they ate for lunch that day, can you imagine if someone had told you that actually in the future that this will be an essential part of creating a anti-pandemic moonshot?

Alpha Lee: Yeah, indeed. Even I would have not expected that. But, yeah.

Eric Ries: Okay. Well, I want to hear, obviously, we want to talk about the moonshot, but let me just back up a little bit. Let me ask you about your background. One of the recurring themes in the people that I've spoken to in the last few weeks is a society wide sudden realization about the importance of science as the engine of progress, And the importance of having science driven policy, as if somehow we didn't know that before, but somehow we didn't. We didn't feel it in that visceral way. You've dedicated your life to science. Talk about why you made that choice. What was it? When you were young, did you have an interest in science? What got you into the scientific field as a career calling?

Alpha Lee: Even as a kid, I was always very interested in how complexity could be built by very simple building blocks, and so curious how nature works in building very complex things. Specifically in chemistry, you often start with very simple building blocks, and then you can rationally build something complex like a drug. That passion started as a kid and carried me throughout school, and I started undergraduate when I was 15, so pretty young, so always very interested in making molecules and thinking about why things work. I think that enthusiasm became crystallized during my undergraduate and graduate training, and realizing science, on one hand, about appreciating the beauty of nature and building complexity, but also about a rational way of thinking and very evidence based approach of making decisions. As scientists, we think about hypothesis, we think about how these hypothesis are translated into theories and how these theory informed experiments, and is that the vibrant, evidence based reasoning process, coupled with the beauty of the objects that it's trying to describe and importance of the problems that it is solving that really keeps me going.

Eric Ries: Talk a little bit about your undergraduate and graduate training.

Alpha Lee: I was a chemist by undergraduate training. I did a lot of organic synthesis to make complicated molecules from simple building blocks. I think of it as being a molecular architect building complex things from simple pixels, simple building blocks. That was a lot of fun. Then after doing the degree in chemistry, I wanted to understand a bit more about the more theoretical aspects of chemistry, so I did a PhD in Mathematics at Oxford, where we focused on taking a more physical mathematical view on science and chemistry. Then for my post-doc at Harvard, I segued into machine learning, where we go even closer to data and really try to extract trends and correlations from looking at large data sets, in particular focusing on drug discovery, which combines my roots in chemistry with my passion in mathematics.

Eric Ries: So then you went on and started a research lab, is that right?

Alpha Lee: Right, so I then started a research group in Cambridge. I still run the research group in Cambridge with 11 post-docs and grad students, where we work on machine learning, algorithms for drug design and molecular design.

Eric Ries: Then somehow you made the transition from the scientist in a lab to the founder of a technology start up.

Alpha Lee: Right.

Eric Ries: Talk about what drove you to do that.

Alpha Lee: We published several key papers on advancing drug discovery, and specifically on how we can fish out signals about which molecules would work or potentially be a drug, even having not enough data, and about a key breakthrough on busy suggesting recipes of how to make complex molecules. This recipe generating process has always been a very determining status, a stumbling block in drug design.

As I was sitting in my lab, I thought academia is a great place to think about very deep ideas, but I really want to make sure these ideas get applied in the industry so that drugs get discovered faster and more patients can benefit from this technology, and realized that a very effective way of doing that is via starting a company and really collaborating at a very deep level with biotechs and pharma. We're very fortunate that the YCombinator program took us on and is our incubator company, all of that.

Eric Ries: When was that? When did you apply to YCombinator?

Alpha Lee: YCombinator reached out after one of the key papers got published.

Eric Ries: They reached out to you?

Alpha Lee: Yeah. One of their partners reached out to us after the key paper got published late last year. We already had the idea in my mind of starting something, and that really crystallized the company, and then we joined the winter 2020 batch, so we arrived in the States in January this year.

Eric Ries: So, you were actually in the YCombinator program as this news was breaking?

Alpha Lee: Yes, yes.

Eric Ries: What was that like?

Alpha Lee: From a YC prospective, obviously that changes dynamics. We go all virtual. The demo day, for example, is completely virtual, was completely virtual.

Eric Ries: First ever?

Alpha Lee: Yeah, first ever virtual demo day. It changes the mechanics, but I think the community is still very strong, and at the end of the day, it's the community of founders, the very strong mentorship of the YC partners that really makes the program a great success, and I think that's still there.

Eric Ries: Let's talk about drug discovery and how it works. This is a topic that used to be considered arcane, and now the whole world wants to learn more about, and we all have a personal stake in understanding. Start with just how did you become an expert in the part of the drug discovery pipeline that you have specialized in, and explain a little bit about how that part fits in to the overall drug discovery process.

Alpha Lee: There are two, I think, parts in drug discovery: the preclinical stage, and the clinical stage. I focus mostly on the preclinical stage. Clinical stage pertains to when the drug actually gets into humans and you test with it. it's efficacious in humans. The preclinical stage is everything before that. Within the preclinical stage, you basically need to find a protein, find a machinery in the body, or in this case, in the virus, that you want to target.

Eric Ries: We're talking about biological machinery.

Alpha Lee: Biological machinery, yes. In this case, we pick the main proteinase, which is a machinery for bio-replication. Then you will need to design chemical molecules that can either stop the machinery from working, or in other human diseases, promote the machinery. In this case, you want to stop the machinery so the virus cannot replicate. Designing molecules, and think of this as drawing out the blueprint to build a skyscraper, but then the second stage is actually building that skyscraper, that molecule. Now, that make stage is actually one of the great determining steps in drug discovery. I can draw complicated things, but I can't really make it. That's where we innovate a lot as a research group, and also as PostEra, the company, how to accelerate this process.

Eric Ries: So, within the preclinical stage, we're talking about designing and then making molecules that can promote or disrupt some part of that biological machinery. PostEra and your research is focused on the building side of the preclinical stage, is that right?

Alpha Lee: Right. We focus on accelerating the building side, and also designing complex molecules that can be built. So, we focus on how to make molecules a lot faster by suggesting a recipe, but also coming up with designs that can be rightfully made.

Eric Ries: How much of the difficulty in drug discovery is... many of our listeners will have taken an organic chemistry class once upon a time, are we literally talking about just coming up with a molecular sequence that can have a certain property, and how much of the difficulty is once you've found something that at least in theory should have the effect that you want, actually figuring out how to make it in the real world?

Alpha Lee: Could be a very challenging question. For example, just designing recipes to make, let's say, 2,000 molecules, which is the scale of designs that our open source drug discovery project, COVID moonshot, received, would take a chemist weeks to just come up with these recipes for thousands of molecules, whereas a computer can do that within a weekend. That's the scale of complexity once you're dealing with building complex molecules.

Eric Ries: Give us an example of what you mean by a recipe, by what you mean by a blueprint. Help us understand the different elements of the stages. I think that will help people understand why this technology is potentially revolutionary.

Alpha Lee: Right. A blueprint is saying, "This is the molecular structure that I want to make, and this structure satisfies the rules of chemistry. I can visualize on the computer, but I can't really hold the structure in my hand or even have a bucket of this structure." By recipe, I mean, "Okay, the way I make this complex structure is by starting from simple building blocks." Some of them derived from oil, some of them derived from other natural products. These are very small, simple building blocks. Then through the machinery of chemistry, you join these building blocks together bit by bit to build this huge building. So, if you think of building a skyscraper starting from really simple architectural materials.

Eric Ries: What are the materials in this case? Give us some examples.

Alpha Lee: Those could be very simple building blocks, like benzene or toluene, or benzoic acids, or the route of some of these. So, basic things that you can extract, let's say, from all petrol, chemical, oil, and then refine and functionalize the forms of these very simple chemistries.

Eric Ries: Do you happen to know a story about a past drug, like its drug discovery story, how many different trials or candidates? I'm just wondering if there's a story like that that you could tell that would illustrate the conventional approach.

Alpha Lee: Conventionally, a drug discovery will be a multi year process, so six years or three, four years would not be outrageous in the pre-clinical chemistry stage of drug discovery. You will typically make a lot of compounds, so 3,000, 4,000 compounds. If you don't accelerate the process, then on a per compound basis, the time that you are potentially prolonging due to not having machine learning compounds relatively quickly when you're dealing with thousands of compounds.

Eric Ries: So, conventionally, if I have a drug on my drug shelf that I got a prescription to that was developed 10 years ago, 15 years ago or longer ago, for every drug that we use now, somebody had to do this drug discovery process.
Alpha Lee: Oh, yes.

Eric Ries: We're talking about thousands of candidate compounds that had to be tested over the period of two, three, four, five, six years just to find the one that was worthy of going to clinical trials, and the drugs that we take as members of the public now are the survivors of that tournament system, the ones that survived those thousands of candidates in the pre-clinical trial, and then in a clinical trial, were shown to be safe and efficacious.

Alpha Lee: Yes, absolutely.

Eric Ries: When we're talking about using machine learning here, we're talking about taking the description of a molecule, you'd like to be able to have a list of the available ingredients you could make it out of, and the chemical transformations that you can do in the laboratory, and trying to discover a pathway from the materials that you have at hand to the compound in question.

Alpha Lee: Exactly. That task itself is a very challenging one for human chemists to think about--not very that many different chemical transformations that one can think about executing, and which order you want to secure them in is difficult to determine. There are so many building blocks, small building blocks that are available, more than a billion small building blocks that you can buy.

Eric Ries: So, it's an unbelievably complicated and large search base.

Alpha Lee: It's an unbelievably complicated search base.

Eric Ries: Talk a little bit about how this used to be done pre-machine learning. We would literally be talking about a chemist at the whiteboard or using their imagination trying to cut down on this search base somehow, and that's why it was so time consuming.

Alpha Lee: Indeed. A chemist would draw a molecule on the whiteboard and say, "Hey, let me cut this bond," and that becomes two smaller, less complex molecules, and then, "Okay, from the less complex molecule, I can cut another bond and form even less complex molecules." Then, "Oh, this less complex molecule turns out to be something I can buy." Then, great, let's spring to action. It's very much a trial and error process, not to mention that even a scheme that a human chemist comes up with may not be actually executed in a lab, because you have imperfect information on which reaction will work, and our algorithm resolves that by basically looking at the entire literature, chemistry literature at least published in patents, and learn the certain rules of chemistry from scratch.

Eric Ries: Wow. A lot of the discussion and news around this pandemic has focused on the timelines to get therapeutics and vaccines to market, and I think a lot of the discussion has been about the very lengthy human trials, the clinical stage of drug discovery. So, help us understand both with conventional approaches and then if this moonshot were to work, what could these approaches mean in terms of that timeline? People, I think, are thinking, "Okay, well, 18 months or 24 months, how could that be accelerated?" Mostly, the conversation that I've been privy to has been about accelerating human trials or figuring out how to evaluate safety and efficacy in a more efficient way.

In fact, on an earlier episode, we talked about the vaccine Manhattan project, which is focused on making the safety trials for vaccines go a lot faster. This is a completely different part of that timeline. For a drug that would take 18, 24 months to come out, how much of that is taken up by drug discovery?

Alpha Lee: I will say that we accelerate the drug discovery cycle by making sure that we make test cycle runs as smoothly as possible. I've talked about the design and the make. The make is the time drainer. It takes weeks to make a compound, so shaving off that accelerates the whole cycle. The design, we save a lot of time by opening up to the public, and leveraging with some of the crowd in conjunction with our ML algorithm, and the test phase is basically just testing the molecule against the COVID protein and also COVID virus. That part is already pretty routine, and we use some algorithms to speed it up, but I think the major saving here is opening up the design funnel so that we get as many great ideas coming into the pipeline as possible, and that's why we have this idea of opening up crowdsourcing, and then really shaving off time in the make stage, the build stage, and then the test stages about designing the best experiments, the most informative.

Eric Ries: Out of curiosity, do you test against live virus or a pseudo virus? How do you do the testing?

Alpha Lee: Right now, we are testing against the protein, because that test part is a lot higher throughput if you're doing it against proteins than against viruses, live viruses, which cause access to secure labs. But we have aligned quite a few virologists around the world who would be very happy to test promising candidates against live viruses, and even animal models of infections.

Eric Ries: Talk a little bit about the iterative nature of this, because I think the other thing that is confusing to the lay public is shaving a few weeks off drug discovery doesn't sound very significant if you imagined we're only going to do it once, and then, okay, in an 18-month drug discovery process, a few weeks, sure, it matters, but it doesn't seem that dramatic. But talk about the benefits of being able to run through this iteration cycle much faster.

Alpha Lee: Typically, these cycles are not one shot and you discover the best drug immediately. Typically, you run through several design cycles where the make stage could be a month to two months, so if you can shave it to a few weeks, that would help a lot, compounding all of your cycles. In the design stage, if you have diverse ideas driven by machine learning to make sure that these ideas are iterative and improve upon the random search and chemical space, means that you can reduce the number of cycles to hit the desired molecules, which means that you decrease cycle time by making sure that you make molecules faster, and you decrease the number of cycles by making sure that you are using crowdsourcing and diverse inputs to go into the cycle, but machine learning to prioritize the molecules you make so that you're not mindlessly searching chemical space, but there's a very defined search direction.

Eric Ries: Usually, when we talk about wisdom of crowds and machine learning and these kinds of approaches, among the benefits are that there are ideas out there, hypotheses, candidates that are too crazy or wacky to be tried in a conventional process, which is by nature more conservative about conservative resources and can't run enough experiments to really try the crazy stuff. Of course, in the crazy stuff is sometimes where the breakthroughs lie. Is that dynamic at play here as well?

Alpha Lee: Absolutely. We released structures of these very, very small molecular fragments, so-called fragments, sitting in the protein. There are many, many ways by which you can stitch these fragments together to form a full fledged molecule. Machine learning provides some insights, but there's a lot of medicinal chemistry ideas of creativity, or some would say even witchcraft, how they stitch these bits of molecules together. Just by releasing it to the public and getting a lot of medicinal chemists from both industry, academia, retired, medicinal chemists chiming in, you get a lot of great design ideas. More than that, we operate also a forum where a lot of the scientists can discuss ideas, discuss strategies, which means they gather a lot of interesting discussions on how to pursue the chemistry or how do we improve the molecular properties, et cetera.

I think we are getting a lot of this going on right now, which means the project can be massively accelerated. It's like having the best chemists from around the world all sitting in one virtual room talking about chemistry. That's even unheard of in the industry.

Eric Ries: Talk about the founding of PostEra. Talk about pandemic struck, what you thought the company would do and what the early progress you had made before you made this conversion over to the moonshot.

Alpha Lee: PostEra offers medicinal chemistry as a service, so powered by machine learning, where we help pharma and biotechs accelerate their drug discovery cycle by helping them design better compounds, make compounds crucially faster, and then informing them which experiments are the much informative to completely close the design, make, test cycle. A lot of AI and drug discovery is certainly not a new area, per se, but we took a step back and realized that the compound making part is a key pain point in the whole cycle, and that, coupled with experiment design, how do we design experiments, how do we know what we don't know, so the design experiment is most efficiently, drives the whole thesis of the company.

We've got a lot of traction as a company. We've got several biotech clients interested in our product, willing to collaborate with us, a project with the American Chemical Society on benchmarking machine learning models, offering insights on how to evaluate machine learning models. So, yeah, it was going really well. We still think that this is a very important area to write for disruption. Obviously then moonshot came, which we are very happy to lead pro bono.

Eric Ries: You talked about having that epiphany one day when you realized that you saw this Tweet online, you started interacting with a group of research scientists that the technology that you were working on for the company could be useful for COVID therapeutics. What did you do the next day?

Alpha Lee: Well, after we Tweeted at Oxford and they replied, we immediately launched into a Zoom call, where we then discussed, "Okay, we are this far into the technology. How do we bring the science and technology together?" We immediately thought, "Well, the best way is to open up the community so that we get as many ideas as possible going to the funnel, and then we pluck our design, make, test platform in the middle of this workflow, and so the next day we just basically launched a website, COVID moonshot.

Eric Ries: Literally the next day?

Alpha Lee: Literally two days later, we launched a website, primarily a website with compound sketching tools, et cetera.

Eric Ries: That has been such a theme of these conversations, how many of the projects that have had impact during these crisis were launched in two days or less.

Alpha Lee: Yeah, literally. Once we knew that we should do it, we just did it.

Eric Ries: Did you have any pushback from your investors or anyone else on your team or in your extended ecosystem? What was it like to tell them, "Hey, we're putting our whole startup on hold and we're just going to do nothing but COVID relief now."

Alpha Lee: They have been overwhelmingly supportive. I think all investors know that COVID is a once in a generation crisis, hopefully. Hopefully we won't have another one of these.

Eric Ries: So we pray.

Alpha Lee: Until many, many years later, if ever. I think we realized this is the moment where everyone should work together, and I think we got overwhelming support. I think also through working in COVID, we are offering this technology and obviously bettering the technology as well so that we can better serve commercial clients later on.

Eric Ries: So, you put the website up and invited the world to join. It was an open source concept really from the beginning, right?

Alpha Lee: Yep.

Eric Ries: That's a technique and a form of social engineering, almost, that is usually thought of as coming from the software world. It's not so commonly applied in the physical sciences and the biological sciences. What inspired you to take that open source approach?

Alpha Lee: I think it's a meta pragmatism, because we know that we need as many inputs as possible, and open source is a great way to encourage input, and also in anti-infective therapeutic areas or antibiotics, antiviral, even before, I hope not after COVID, but who knows, but certainly before COVID, this area has not really received much investment by big pharma.

Eric Ries: Shockingly.

Alpha Lee: For the very simple reason that developing a drug which can only be used once every 20 years is not something that is particularly attractive. So, we thought the only way to achieve short run momentum and long run success is to make sure that we have a community behind this effort, and make sure that all the data's available so that obviously we want to discover a drug, but even if we can't quite reach a clinical drug, at least all the data and all the insights are there so that when the next pandemic strikes, there is a resource for the world to lean on and to think about.

Eric Ries: I really want to thank you for taking that approach. I think that's so sorely needed, and the need for collaboration and preserving and sharing data has never been more evident, and so taking that long term view, whatever happens next, I think is a commendable thing to have done. So, what kind of initial response did you get? You put this website up, you made this call for scientists around the world to send you their ideas for what could be tested in this drug discovery process. So, what kind of response did you get?

Alpha Lee: We thought we would get maybe 100 submissions, max. But instead, we got now over 4,000 submissions from 280 chemists around the world.

Eric Ries: Who are these people? That's just a remarkable thing. I would have thought, "Gosh, how many chemists can there even be with the knowledge to be able to do this?"

Alpha Lee: Yeah.

Eric Ries: Who are they? Who are they? We're not talking about kids on their couch, right? When we talk about crowdfunding, who are these people?

Alpha Lee: They are a lot of chemists working in industry, a lot of retired chemists. In fact, several of the frequent submitters joining the core group of moonshot, they are ex- Pfizer chemists, recently retired. We get a lot of academics, medicinal chemists, obviously working from home. Class is canceled, so they have time and creative energy to contribute. I think once we lean on the community, the community really has uplifted us to an even more promising stage.

Eric Ries: What were those initial submissions like?

Alpha Lee: They were great. They were basically meticulously joining these breadcrumbs, following the breadcrumbs, joining the molecules together to form potential drugs. I think the initial submissions are amazing. They really creatively designed around the ideas or the experimental evidence that's out there.

Eric Ries: What fraction do you think of the ideas that you have tested would not have seen the light of day if not for the moonshot?

Alpha Lee: I think we would have tested different ideas. For example, you can just make a lot of molecules and just test them, rather than having a more intellectual way of selecting which molecules to make and test, which means that the success rate would be significantly lower. That means you get less promising molecules to go to the next cycle, and the less promising molecules will therefore have a low probability of success at the end of the day, which means that the project is much less likely to get approved.

Eric Ries: I guess what I'm asking, let me try to ask a different way.

Alpha Lee: Okay.

Eric Ries: Because I think people have heard about... I can't pronounce any of these drugs, but Chloroquine and Remdesivir, is that right?

Alpha Lee: Yep.

Eric Ries: Some people have heard about these candidate drugs. We did a previous episode entirely about niclosamide, and I think what people will want to try to understand here is how many of the candidates that you're even evaluating are drugs that would otherwise have eventually been tested otherwise, if not for this project, and how many of these are genuinely new ideas that wouldn't have otherwise seen the light of day.

Alpha Lee: All of the ideas are genuinely new.

Eric Ries: Yeah, so explain that, because I think that's hard to imagine that thanks to this almost 5,000 drug ideas are going to be tested that would otherwise have not been.

Alpha Lee: Right.

Eric Ries: It seems like a lot.

Alpha Lee: I think a lot of the approaches that are currently being done are known as repurposing. The idea is that you take drugs that are already available in the market past FDA approval for other diseases, and then you test them against the virus and hope that one of these candidates will successfully kill the virus. There are only, I think, around about 12,000 of these repurposeable molecules out there to be tested. Remdesivir came from one of these campaigns. It shows an antiviral, but for another virus, Ebola, originally developed for, whereas our approach is genuinely designing new molecules, starting from scratch, so all the ideas that people come up with, not something that you would usually test against COVID, because these are not really drugs, these are ideas, genuine chemical ideas.

Eric Ries: It's like a factory for drug innovation.

Alpha Lee: Exactly.

Eric Ries: You got this initial wave of submissions. Bring us up to date what's happened since.

Alpha Lee: We use our ML technology to accelerate synthesis of these molecules, making these molecules, and we got a few hundred back now from the synthesis lab, and we've tested a few hundred. We've found more than a dozen very strong hits, so molecules that can stop the protein from working.

Eric Ries: We're talking about the spike protein in coronavirus, right?

Alpha Lee: The main proteinase. The main protein is a machinery that the virus relies on in replication. When the main proteinase gets inhibited, the viral replication becomes difficult, and the virus would therefore die.

Eric Ries: I see, so just make sure I'm understanding this, because we had talked about the spike protein in previous episodes, about how the virus infects existing cells and enters into the cell's machinery.

Alpha Lee: Exactly.

Eric Ries: This is a different protein that is used to replicate the virus once that penetration has taken place.

Alpha Lee: It's a protein for virus replication.

Eric Ries: And if the virus can't replicate, it's no longer dangerous.

Alpha Lee: Indeed. Proteinase are proteins that break down other proteins. Main proteins in particular is a machinery that the virus uses for replication. The drugs we're designing are called inhibitors, i.e the function is basically gumming up and clogging this molecular machine and stop it from working.

Eric Ries: If one of these candidates that you're currently testing proves to be viable against the virus, what's the timeline from here?

Alpha Lee: It would take, I think, around six months to get to a so-called preclinical candidate, sort out all the chemistry part, now we have a optimized molecule, looks great in virus, looks great in the protein, looks great in clinical animal studies, and then it would take a few more months to do the more intensive studies to determine the toxicity of the molecule in animal models, scale up the molecule in a safe way without impurities, et cetera, and then after that, in human. It's around a year.

Eric Ries: Are there more steps that could be taken to accelerate that timeline if the need proves greater than we currently anticipate?

Alpha Lee: I think what will be needed is much more funding in novel drug discovery, like moonshot. Right now, we all pull together on a pro bono basis, but to accelerate that, we need significantly more funding so that we can make more molecules and really accelerate the testing cycle.

Eric Ries: Who needs to do that? Are we talking about scales of funding that only the government can do, or is there a role for private philanthropists and non profits here? Where does this funding need to come from?

Alpha Lee: Yeah. I think private individuals or grant agencies will be, I think, particularly needed. We have actually launched a GoFundMe page to crowdsource the funding, and we have now reached over 10K in funding. Our goal there is 2 million. I think we will definitely be very grateful for support from private individuals. I think states as well should think about funding novel drug discovery. I think a lot of the pushback that I've seen and heard is over drug discovery takes time. Why not just repurpose drugs or vaccines or whatnot? I think all these repurposing vaccines are very important efforts, but a novel drug discovery is crucial, because if repurposing doesn't work, if a vaccine doesn't work, then this is the last shot at coronavirus, so we need to have this.

Eric Ries: Yeah. I think we can do a lot better than $10,000, even just among the listeners of this show, and I'll pledge as well. We'll put a link to the Go Fund me campaign in the show notes. But I think this has been a theme over and over again, is as a society, we don't want to make the longterm investments. We just want the quick wins. The drugs we repurpose today are the ones that happen to have gotten through this discovery process in the past. So, if we hadn't made those investments in the past, we wouldn't even have any drugs to repurpose.

The discourse right now is about how a vaccine and a therapeutic, they're sure to come. Everyone's got these timelines in mind that we can reopen the economy, but that is all predicated on the understanding that the virus won't mutate, that there won't be a second strain, that these approaches could work. Not to paint a bleak picture, but what if we're wrong? What if a vaccine is more difficult than we currently anticipate? What if the currently available therapeutics are not effective or the virus develops a resistance to them? This is an all of the above moment where we have to be putting our money where our mouth is. We say that it's important to get the economy open. Well, here's a chance to make that a reality. I don't know about you, but doesn't $2 million seem incredibly cheap, even if all we do is get the American economy reopened eight days sooner?

Alpha Lee: Yeah. I think it's a great investment. To be even more provocative, 17 years ago, SARS-CoV-1 struck the far east and paralyzed many cities there. Growing up in Hong Kong, I experienced that firsthand, the paralyzation of the entire city. The main proteinase of SARS-CoV-1 and SARS-CoV-2 shares a 96% similarity, which means that if the world had persisted in discovering a proteinase inhibitor against the 2003 version of SARS, COVID-19 might not even be a pandemic.

Eric Ries: Yeah.

Alpha Lee: There might even be a cure. Instead, once the pandemic subsided, the world decided to move on and look the other way. I think it is very important that drug discovery efforts continue, because even if it's not COVID-19, there's another coronavirus out there. If the world only looks for short term solutions, then you will always get pandemics over and over again.

Eric Ries: I wish we could put that on repeat and just blast it out of speakers all across this country. That's exactly it. This epidemic is partly a virus, but it's also an epidemic of short term thinking.

Alpha Lee: Yeah.

Eric Ries: If the world leaders who were on the stage in 2003 had taken this seriously and made the investments, we're talking about how to accelerate these drugs from 18 months to 12 months, from 12 months to 6 months, but we had 17 years to do this. Our commitment to basic research, to drug discovery, to these fundamentals is so inexpensive compared to the cost of the economic damage that these pandemics can reach, and nobody can say they weren't warned or they didn't know. Epidemiologists and scientists have been sounding the alarm about this for years and decades. We, the general public, we, the literate general public in particular, have embraced complacency and the quick sugar rush of short term thinking, because it's been easy to do so.

We have to, for our sake, for our children's sake, we have to get serious about this and make sure that we never return to that state of complacency, and that we make the investments. Right now, today, there's a GoFundMe page that you, if you're listening to this, you could be investing in the drug that will allow us to leave our homes in short order. But also, that we hold our policy makers accountable, that we hold our leaders accountable in not just in government, in academic, in industry, any of those institutions. Any institution in the world could have been keeping this research alive this whole time, and all chose not to. That needs to be our collective shame and a call to collective action going forward here.

Alpha Lee: Yep, completely agree.

Eric Ries: What is the rate limiting step though in scaling this up so that we are at the scale of mitigation that is needed for this crisis?

Alpha Lee: I think resources in drug discovery efforts is the key step, and making sure the resources are continuing to exist after the immediate pandemic, so we don't forget. I think that's an important point. I think from an even broader perspective, how to make sure that the policies about funding and policies about pandemics are driven by evidence. I think that is possibly a more long term discussion, how to make sure that social policies are driven by science.

Eric Ries: Amen to that. If we haven't learned the lesson of the importance of science driven policy now, I don't know that we ever will, and I don't know that our civilization stands much chance of surviving if we don't. What are the other organizations? If you were talking to a policy maker, a state or a national leader now, urging them to make investments that could both help now and plant the seeds for future recovery beyond the moonshot, where would you be urging them to send resources?

Alpha Lee: I think the ideal state would be a state run organization, research organization just focusing on anti-infectives, where these are obviously assets that society needs, weapons that we need to fight against pathogens, both bacteria and viruses, yet it is clear that it is very difficult, and understandably so, for industrial players to participate. I think the state should just shoulder that, and have a team to discover molecules against these viruses or pathogens, even if they will never be used. Just stockpile the weapons against the pandemic. A lot of people have advocated stockpiling or designing antibiotics and other anti-infectives. I think the state should actually do something, and that's a massive undertaking, obviously, because drug discovery efforts are expensive.

Eric Ries: Well, not compared to maintaining a nuclear arsenal.

Alpha Lee: Right.

Eric Ries: Which we do on a preventative basis, just in case it proves necessary, so I don't think their proposal is radical at all. Have you given some thought to how much it would cost to fund such a program?

Alpha Lee: The funding the program I discussed, I talked about, was for the preclinical stage of drug discovery, because for this pandemic, I think we can think of once we have the molecule ready, we hope that industry or state grants would allow us to take this to clinic. To give you clinic phase one, phase two, phase three, is another hundreds of million endeavor, and if the state would establish basically a pharma focusing on anti-infective, you're dealing with a few billion of investment, at least. But still, that pales in comparison to the human lives that are lost in the pandemic, and obviously economic damage the pandemic causes.

Eric Ries: Yeah, even a billion dollars a year in perpetuity is really quite modest compared to the human and economic toll.

Alpha Lee: Yeah. I think the United States is a very strong actor to lead that, to start this effort. The US has a very strong record of drug discovery, of innovation in science. I would advocate to any sort of policy makers who might be tuning in to think about that.

Eric Ries: What about the changes that need to happen on the private sector side? Because it's funny you say that, we have had this leadership position in the world economy in the past, and yet the trend among all public companies, this is not unique to pharma, but pharma included, has been to deconstruct their R&D efforts to outsource drug discovery and other kind of basic research, R&D labs to third parties, to other countries, to startups. So, our overall innovation capacity in the private sector has actually declined, even in a period where our material prosperity has increased. What are the changes you think need to happen within the private sector to the pharma industry and to our capital markets, more broadly, as a result of this pandemic?

Alpha Lee: I think pharma, there's a lot of worry about pharma losing productivity. I think the productivity, firstly, I think a lot of productivity has been shifted to biotechs, which still stays within the economy. I think biotechs being able to create new ideas that would then get sold to pharma, I think it's also an equally valid pipeline. I think what is losing investment is specific therapeutic areas like anti-infectives, where, see, the collapse of Achaogen for example, one of the few companies with new antibiotics on the market, typically when you have a new drug in the market, you celebrate this by a huge shot in the stock price and whatnot, but instead, the company collapsed shortly after the drug was approved, and that's unheard of in any other therapeutic area. I think it's these cases which shows how capital is being fairly inefficiently allocated.

Eric Ries: Do you have any books or other web resources you want to plug, or that you think people should pay attention to, things you want us to link to, anything like that?

Alpha Lee: I think the GoFundMe would be quite nice, yeah, because we are looking for funding right now, and we are basically 30 chemists in research labs in Ukraine, China, India, working for us making molecules. The current burn rate wouldn't last us very long, and we are committing all our resources now because we think that we don't do it now, we don't do it ever. So, we really need funding ASAP. Yeah. The GoFundMe would be great.

Eric Ries: Tell me more about the GoFundMe campaign and what the money will go towards.

Alpha Lee: The Go Fund Me campaign is about raising resources for making and testing these molecules. PostEra, although we are organizing this, this is not in any way tied to PostEra's commercial operations. It's solely used to make and test compounds against COVID. Every hundred dollars will allow us to test one more compound and hence one more shot at finding a safe and efficacious antiviral.

Eric Ries: Is this a non-profit? Do you have a non-profit that is sponsoring the funds or anything like that?

Alpha Lee: We are organizing Go Fund Me, but the donation can also go to universities, so Oxford, Cambridge, or Weizmann. For those who want to make tax deductible donations, there are universities behind us, Cambridge, Oxford, and Weizmann Institute, and they will be able to accept tax deductible donations.

Eric Ries: We'll put more information about this campaign. I just can't think of a better use of $500 right now than to tackle another drug. I mean, the one that you sponsor could actually be the one that gets us out of lock down someday. It seems like even though the probability is small, the payoff is so immense as to be worth it. Talk a little bit about, we didn't really talk about how you and your co-founders are stuck here in Santa Clara doing this. Just talk a little bit about your living situation and how you're trying to save the world from your Airbnb.

Alpha Lee: Right. We are staying in the same Airbnb that we stayed in the past five months since arriving in California for YC. Yeah, three of us, three co-founders, we all stay in one apartment. That actually makes working with this very easy.

Eric Ries: But hard to get a break, I guess.

Alpha Lee: Yeah, but I think we're all very excited about it, so it's very good to be just cranking it for the challenge.

Eric Ries: Far from your family though, yeah?

Alpha Lee: Yeah, yeah.

Eric Ries: Has that been hard?

Alpha Lee: It's been okay, I think. Obviously I think being through the SARS pandemic when I was a kid back in Hong Kong, I was seeing this being played back again. Same deal, schools, society grinding to a halt, cities get temporarily wrecked. It's like deja vu. It seems that we have not learned too much. That's a very saddening thought.

Eric Ries: Does it frustrate you that the world didn't learn from the sacrifices that you made in Hong Kong back then?

Alpha Lee: It does seem that there's a missed opportunity. I wouldn't say frustrated, because I think, for example, a lot of cities around the world have used the response strategies of the far east as a template for how it should respond to pandemics. But I think on a broader scale, how drug discovery efforts get dropped, for example, where you see investment anti-infectives, the investment, they have not only not increased, but actually decreased over the past decade and a bit. I think that's pretty disheartening.

Eric Ries: What's been keeping you optimistic during this time?

Alpha Lee: I think what's keeping me optimistic is that the community, scientific community and community in general is really showing the best of itself. So, coming together in the moonshot, which is a small part of the community, obviously, in addition to obviously doing a lot more other things, vaccine development, repurposing. All the scientific groups are all chipping in. I think that's really promising, and obviously society and community at large, how we see a lot of individual acts of kindness are being displayed, I think that shows that there's a lot of hope in humanity.

Eric Ries: Where do we go from here? How do we get out of the crisis?

Alpha Lee: I think the sooner we subscribe to evidence based policy making, the quicker we're going to get out of the crisis. I think there are obviously a lot of decisions that inevitably engage in things that are much more than just numbers. There is a genuine policy or political philosophy dimension to a lot of the decisions. These are why we have politicians being elected to do the job for us. However, I think all of these decisions should be based on a shared set of evidence. Based on the evidence, different people may have different takes, but one should not disregard the evidence or try to twist the evidence in a way that supports a certain outcome.

Eric Ries: Amen to that. Alpha, I want to thank you for the work that you're doing that benefits so many of us and that could be critical to ending this nightmare that we're living through. Of course, thank you for taking the time to share the story with me. I really appreciate it.

Alpha Lee: Thanks so much, Eric, for having me.

Eric Ries: It's really been a pleasure. Let me pledge my support, and I think I can speak for a lot of people in the community, that we're standing by to help. If at any point you need volunteers, you need help, you need software engineers, you need designers, you need anybody, apart from those who work in the lab, those of us who are civilians are standing by to lend our assistance.

Alpha Lee: It's greatly appreciated. Thanks.

Eric Ries: Thank you so much. This has been Out of the Crisis. I'm Eric Ries. Out of the Crisis is produced by Ben Erlich, edited by Jacob Tender, and Sean Maguire.  Music composed and performed by Cody Martin, hosting by Breaker. For more information on the COVID-19 crisis and ways you can help, visit If you are working on a project related to the pandemic, please reach out to me on Twitter. I'm @ericries. Thanks for listening.

blog comments powered by Disqus