Over 10 years experience of Traceability Solutions
By pharmatrax
Category: Technoloy
No Comments
“Follow the money,” you’ll often hear people say, and that small bit of advice is why we pay close attention to who invests in particular companies and technologies. Firms that have lots of money to invest in risky ventures are likely to go through the difficult due diligence process to validate someone’s technology, intellectual property, and capabilities to execute on a vision. The same logic holds true for identifying talent.
Take the recent news about a startup called Insitro landing a deal with Gilead to use machine learning to find therapies for NonAlcoholic SteatoHepatitis (NASH). The person behind Insitro is Daphene Koller, a Stanford University professor and longtime machine learning researcher who co-founded Coursera alongside Andrew Ng who recently raised capital from Intel for a new venture. Ms. Koller has attracted investments from the likes of Google and their subsidiary Verily, hedge fund Two Sigma, and Jeff Bezos. A partnership with Gilead is icing on the cake, and it’s just another example of progress being made in using machine learning for drug development.
We recently wrote about a handful of new firms that are using computational chemistry to find new drugs. As we noted, computational chemistry is a subset of computational drug discovery. In order to better understand what’s happening in the broader world of computational drug discovery, we popped into the Mountain View office of twoXAR to speak with co-founder Andrew M. Radin who has his thumb on the pulse of machine learning and drug discovery. He’s been working a lot with Korean firms lately, and we were curious to know why.
In our recent series on the global AI race, one of the countries we looked at was Korea. In that article, we noted how the Korean government’s aim is “to join the global top four nations in AI capabilities by 2022 with the establishment of at least six new schools focusing on AI and the training of more than 5,000 engineers.” Since that article was published, the Korean government announced their intent to leverage AI for drug development and boost their companies to the top-50 global pharmaceutical company list.
Every year, there’s a list published by PharmExec.com showing the top-50 pharma companies by revenues. If you don’t work in the industry and know the big names off hand, it’s useful to know who sits at the top and what top drugs they’re selling.
When looking at country representation across the list of 50, in first place you have the U.S. (16 companies), then Japan (10 companies), and in third place zee Germans (5 companies). One country that you won’t find on the top-50 is Korea, and if you’ve been paying attention to the news surrounding machine learning and drug discovery, you might see a disproportional amount of activity taking place in Korea.
We noted one AI drug discovery startup in our Korea top-10 AI startups list – Standigm – which “currently has more than 30 therapies in development, including five NASH therapies in animal trials.” Several Korean pharma companies – SK Biopharmaceuticals and Daewoong – have announced their own AI programs. And Korean firms have been reaching out to companies abroad looking for deals.
Mr. Radin has been spending quite a bit of time in Korea, and that’s where he jetted off to after we left his office in Silicon Valley. Late last year, he attended the AI Pharma Korea Conference in Seoul where various companies gathered together to talk about what direction Korea needs to take to reach their goals of becoming a leader in AI drug development. Right now, the innovation is coming from small firms like twoXAR, and they’ve already inked two deals with Korean pharma companies with one announced just today.
That latest deal with SK Biopharmaceuticals is twoXAR’s fifth publicly announced deal which bears a shared-risk, shared-reward structure. Mr. Radin talked about how each potential drug candidate represents a program that’s centered around one particular disease. The end result is a portfolio of drug candidates that provides the same sort of diversification benefits that a large portfolio of stocks would. We’ve talked before about how the evolution of business models is a key characteristic of emerging technologies, and it seems like the twoXAR business model helps reduce risk and the volatility of future revenues.
Another interesting thing we noted was that of the other three deals twoXAR has inked, two have been with Japanese firms – Ono Pharmaceutical and Santen. Perhaps that’s a sign that Japan companies are concerned that the only way they’ll be able to maintain their dominant position in the global top-50 list is by pursuing AI drug development. While most pharma firms prefer to get involved later in the development game, the Japanese appear to be getting started much earlier. Countries like Japan and Korea clearly recognize the competitive advantage computational drug discovery brings to the table. What about the United States?
We talked earlier about “following the money,” and that isn’t just about who is investing in computational drug discovery startups. It’s also about deals that get signed in this space. Which one of the top-10 pharma companies around the world will be the first to receive approval for a drug that’s been discovered by an AI algorithm? How long will it be until all new drugs are discovered using AI? Mr. Radin talked about how the traditional drug discovery process is so costly that pharma companies can barely realize a positive return on investment when it comes to developing a new drug from the ground up. That’s the motivation for any big pharma company to begin using AI for drug development. However, big pharma players prefer to get involved when the industry sees some real traction. That’s now starting to happen.
During our conversation, Mr. Radin talked about various deals that have been happening which are noteworthy in that they demonstrate some real progress is being made. We’ve talked about all the money that’s pouring into computational chemistry startups, and some of the progress being made by startups like Atomwise who is working with Charles River to offer “AI drug discovery as a service.” Another firm making some headway is Recursion Pharmaceuticals. Incredibly, they’re actually able to look at cells using computer vision to see how they behave when exposed to drugs. Since we last wrote about Recursion, they’ve raised a cumulative $105 million in funding and now have a full drug pipeline in the works.
In July of last year, the FDA granted Recursion an Investigational New Drug (IND) application for a Phase 1 clinical trial of REC-994 in the treatment of cerebral cavernous malformation.
Another company with drug candidates progressing through clinical trials is Boston-based Berg which has completed Phase 1 trials with one compound and Phase 2 trials with another. Their leading product candidate. BPM31510, works by correcting cancer cell metabolism, thereby reactivating apoptosis.
We also saw the first AI drug development IPO last year with BioXcel Therapeutics (BTAI), a company that we had a difficult time understanding. What’s a whole lot easier to understand now is their pipeline which has clearly spelled out milestones and corresponding dates.
Not everyone is making forward momentum though. Just hours ago, Stat released an articletalking about how IBM is “halting sales of Watson AI tool for drug discovery amid sluggish growth.” If you’re a longtime IBM shareholder, you’ve managed to acquire a real tolerance for pain, so this should just be water off a duck’s back. Let’s just go back to talking about how IBM Watson Health Imaging” is the Future of Healthcare.
It’s been a few years since we last wrote about 9 Computational Drug Discovery Startups and things have been moving fast in this space with more than 125 startups claiming to be involved in some aspect of machine learning and drug discovery. The next big milestone will be the first New Drug Application (NDA) for a drug discovered by a machine learning algorithm. The value of AI will quickly become apparent to all, and the cost of bringing new drugs to market will decrease sharply as machine learning becomes a commonly-used tool for optimizing every stage of the drug delivery process.
As the process becomes less costly, pharma companies will be more apt to develop drugs for rare diseases with smaller target markets. Expect to see some consolidation in today’s highly fragmented AI drug discovery market as we begin the downward slide past the peak of expectations. Mr. Radin is looking forward to some of the AI hype blowing off so that he and his team can keep executing on signing more deals and finding more drugs using machine learning algorithms.
Source: https://www.nanalyze.com/2019/04/artificial-intelligence-drug-discovery/