[Interview] How COVID-19 Catalyzed AI-assisted Open Science Drug Discovery
Biopharma companies are now racing to find much-needed cures against SARS-CoV-2, a virus that caused the largest global pandemic of our time. One notable effort is the COVID Moonshot project, organized by an international consortium of scientists from academia, biotechs, contract research organizations (CROs), and pharma -- all working pro bono or via crowdfunding, philanthropy, and grants.
The aim of the project is to rapidly develop easily manufacturable antiviral drugs that can inhibit the SARS-CoV-2 main protease, which is believed to be an Achilles heel of the coronavirus. The project is managed by PostEra, a startup company that uses artificial intelligence (AI) algorithms to map routes for chemical synthesis to speed the drug-discovery process.
After the original crystallography on the SARS-CoV-2 main protease became available from Shanghai Tech University in China in January, researchers at the Diamond Light Source/XChem in Oxford and The Weizmann Institute in Israel performed a fragment screen to identify key entry points for molecular design. Later, PostEra initiated a COVID MoonShot project to crowdsource molecular design ideas from hundreds of scientists worldwide and apply artificial intelligence (AI) to prioritize ideas and find optimal synthesis routes. Chemical synthesis of the promising molecules is carried out by Enamine in Ukraine, Sai Life Sciences in India, and Wuxi in China, while Enamine also coordinates compounds logistics to various screening centers to check them for activity.
The COVID Moonshot is the first open-science drug design project of its kind, and it is an illustrative example of how the COVID-19 crisis stimulates companies to search for new ways to collaborate, using the power of collective thinking and crowdsourcing. Importantly, this project aims to establish a new paradigm for drug discovery, where society’s investment in health would be decoupled from the commercial interests of a single company.
To find out more about the COVID Moonshot project, latest progress, and the role of AI technology in this effort, I reached out to Aaron Morris, Co-founder and CEO at PostEra:
Aaron, what was the main motivation to start the COVID Moonshot project, and why the open science model?
Aaron: Back in March, one of PostEra’s founders saw a tweet from a team of scientists at Diamond Light Source in the UK who had shown that a selection of chemical fragments were effective at binding to a key part of the COVID virus. We realised there were hundreds of chemists sitting at home, with their projects on hold, who could help take these fragments and turn them into genuine drug candidates—an open-science approach to crowdsourcing a new drug. We also realised that PostEra’s synthesis technology could be used to design synthetic routes in a high throughput manner for such compound submissions. Over a weekend we quickly created a platform where designs could be submitted and hoped for maybe 50 to 100 submissions. We’ve now received over 7500 submissions from 350 contributors! For Moonshot the open-science model means we get to tap into the vast array of scientific expertise from people all around the world while also ensuring that IP constraints do not slow down the progress of the project. Additionally, this means we don’t have high R&D costs to recover from the sale of any resulting antiviral and so we hope to produce a cure that is cheap and accessible.
How is the team behind COVID Moonshot planning to sustain the effort financially? Are you planning to reach out to charities, government support?
Aaron: Open science means we are relying on grants and donations to fund our work. We’ve received $20k from our GoFundMe and $290k in grants so far and continue to apply for funding from relevant scientific bodies. Our aim is to fund all the work needed to complete all preclinical studies. We then intend to engage governmental organisations or non-profits to take the drug through necessary human trials.
What’s the latest progress in the project, and what are you focusing on right now?
Aaron: We have now received over 7500 submissions from 350 contributors. We’ve tested 500 compounds and have just broken through the micromolar barrier (top compound has IC50 of 600nM). You can see all the latest activity data on the site. We’ve also set up the assay cascade for in-vivo modelling and will be moving our lead series into both live virus and in-vivo testing in the coming weeks. Our focus right now is on optimising potency further before moving to the aforementioned testing. We continue to release all data via the forum as we receive it.
The current pharma industry business model is clearly failing to address the growing need in such areas as antibiotics discovery, and “niche antivirals” due to a lack of market incentives. Do you think COVID Moonshot can be a model case to ignite a change towards a more “open-science” paradigm to come up with new antibiotics and antivirals?
Aaron: This is definitely something we’ve been exploring outside of COVID Moonshot. Though we are hopeful that new regulation will improve the commercial attractiveness of such therapeutics we realise in the short term that a positive social impact combined with a globally connected community can make headway in these areas. We’ve spoken with leading figures in the Antibiotic resistance fight in the UK as to whether such open science approaches, powered by the Moonshot platform, could be of value to incentivising research into these ideas. We realised that though certainly open-source approaches remove barriers of commercial interest it does produce a new set of challenges -- how to prioritize ideas? How to share data in an accessible way? How to organise the various computational, experimental and research teams? We realised that a committed community, novel technology and a well organised logistical workflow are key to making open-science initiatives work. We also welcome anyone working in the field of Antibiotics or Antivirals who would like to benefit from the type of infrastructure we have built for Moonshot to get in touch and see if we can help your program by leveraging an open-source approach.
What’s the role of artificial technology in the COVID Moonshot project? Do you think AI will be a driving tech in the open science drug discovery movement?
Aaron: Computational methods in general have been very helpful to the COVID Moonshot effort. PostEra’s contribution is to use AI in the area of chemical synthesis. When you have 7500 submissions, how do you decide which compounds to select for synthesis and testing? A natural triaging mechanism is to choose the most readily-makeable compounds. So PostEra used our synthesis algorithms to design synthetic routes for over 2000 compounds in under 48hrs (we expect that would have taken human chemists around 3-4 weeks). This ability to do high-throughput synthesis design has been key to keep the project moving as fast as possible. We’ve also benefited from the algorithms that can search huge spaces of chemical building blocks. Open science efforts will likely always need to work to tighter budgets and so by using AI, not just to design novel compounds but to discover how to make them and where to order them from under budget constraints will certainly help the viability of such efforts.
Topics: Emerging Technologies
There are no comments yet. You can be the first.
Leave a Reply cancel reply
Your email address will not be published. Required fields are marked *