- Alternate Universe
- Posts
- AI Drug Discovery 👩💻💊
AI Drug Discovery 👩💻💊
Owen
November 01, 2024
In partnership with
Welcome to Alternate Universe!
In today’s edition:
Will AI accelerate drug discovery? 💊
Job opportunities at recently funded European startups 👇
The gap between Kering and Hermes continues 👜
We put your money to work
Betterment’s financial experts and automated investing technology are working behind the scenes to make your money hustle while you do whatever you want.
The Crux 🔴
Designing new drugs is hard, really hard.
The often-quoted statistic is that drugs produced through traditional R&D require at least 10 years and over $2bn in funding.
This is a massive problem ripe for disruption.
Thankfully, there might just be a solution - AI-designed drugs.
You already know what time it is.
🐰🕳️⌚
What? 🧐
Until the 1990s, traditional drug design relied heavily on trial and error, limited by the dependence on known molecules. The introduction of de novo drug design marked a shift, enabling scientists to create entirely new drugs with unique properties without relying on existing compounds. However, the approach faced obstacles, especially in synthesizing proposed molecules, which demanded specialized computational skills, constraining its broader application in drug discovery.
The arrival of generative AI in 2017 brought transformative change to de novo drug design, breaking through many of these constraints. These AI-driven models leverage vast datasets on bioactivity, toxicity, and protein structures, accelerating the identification and refinement of potential drug candidates. AI’s promise in drug discovery is considerable—it can make drug development cheaper, faster, and more efficient by digitally handling many of the tedious, repetitive, and costly stages, allowing researchers to explore a wider range of possibilities.
This impact is evident in the growing volume of publications on AI in drug discovery, yet research remains unevenly distributed. AI research predominantly focuses on data-rich, commercially lucrative fields such as oncology and infectious diseases. This trend risks stalling advancements in areas with more limited data, such as mental health, which may be equally deserving of innovation.
Applications 👩💻
How can AI be applied to improve the drug discovery process?
Identification of Compounds
The chemical space - a set of every possible chemical compound - is thought to be in the order of 1060 molecules. AI can reduce research costs by limiting the number of synthesized compounds tested in laboratory settings.
Designing New Drugs
Not every chemical compound has the necessary physical properties to be a viable drug. The Lipinski Rule of 5 is used as a guideline to assess the likelihood of a compound becoming an orally active drug.
Machine learning algorithms can analyze large datasets of chemical structures and their properties to predict molecular weight among other data points. These predictions can help identify compounds that potentially violate the Lipinski Rule.
Drug Repurposing
Using existing drugs, already approved for safety in humans, for new therapeutic uses, enables faster delivery of effective treatments to patients in need.
Biovista uses its AI platform, Project Prodigy, to develop a pipeline of repurposed drug candidates across various disease areas.
Clinical Trial Optimization
AI advancements in identifying and predicting disease biomarkers enable targeted recruitment of patients for Phase II and III clinical trials.
AI Cure is a tool that helps companies understand how patient behavior influences trial results. Its implementation increases adherence to medication in Phase II trials.
So are AI-discovered drugs any good?
A recent review published by BCG tries to answer this question.
The research included an initial analysis of the clinical pipelines of AI-native biotech companies and found that AI-discovered molecules have an 80-90% success rate in Phase I trials. This figure is significantly higher than the industry averages hovering around 50%.
Four out of the ten AI molecules in Phase II trials transitioned to the next phase. This success rate aligns with the industry averages of 30-40%, albeit with a small sample size.
One of these candidates is Recursion’s REC-994, for the treatment of cerebral cavernous malformation (CCM), a potentially fatal condition with limited treatment options. Phase I trial data showed a favorable safety profile. Phase 2 trial data however showed scant evidence of efficacy.
Limitations ⛔
Despite AI’s vast potential, several significant barriers hinder its adoption in drug discovery.
Data limitations top the list. Most datasets are biased toward well-known molecules, which restricts the discovery of entirely new drug targets and limits innovation. Another challenge lies in the synthesizability of AI-generated molecules. While some promising algorithms aim to produce molecules that are easier to synthesize, and new methods to assess synthesizability are emerging, this remains a key hurdle.
Additionally, existing computational tools can be complex or inaccessible, often lacking user-friendly interfaces that would allow broader use. The need for specialized expertise also limits adoption, as effectively applying AI requires knowledge that spans both computational and biochemical domains.
AI-based drug discovery has yet to gain widespread industry acceptance, with many stakeholders skeptical of its near-term impact. The field remains in an exploratory phase, but ongoing advancements may soon address these challenges and bring AI closer to industry-wide integration.
Investment Implications 🤑
The niche has already attracted a lot of VC funding.
Insilico Medicine - has raised over $300m over multiple rounds to develop treatments for cancer and age-related diseases. The company plans to go public on the Hong Kong Stock Exchange.
Cradle Bio - raised a $24m Series A round led by Index Ventures. The biotech company uses AI to help biologists predict protein structure.
Atomwise - has raised over $170m in VC funding to identify and produce small-molecule drugs.
Big Pharma is getting involved by partnering with agile startups. For investors seeking indirect exposure, the key question is: which pharma giants have the most partnerships with advanced AI candidates?
Roche is leading the way. In 2021, it partnered with Recursion in a deal worth up to $150 million upfront and $300 million per program for up to 40 projects, focusing on neuroscience and oncology. In diagnostics, Roche collaborates with PathAI to develop specialized algorithms. AI-driven drug discovery is also central to Roche Venture Fund’s strategy.
Sanofi, another major player, has signed AI deals with Atomwise, Exscientia, and Owkin.
Indirect beneficiaries include data infrastructure providers like Schrodinger and Cradle, which support AI drug discovery. Among public companies, the sector remains volatile. Benevolent AI, for example, has struggled after disappointing Phase II results for its atopic dermatitis candidate, leading to workforce cuts.
One standout is Recursion, whose partnerships and diverse pipeline make it a strong contender. If their CEO’s claims of reducing drug development time to 1-2 years hold, it could be transformative—though the company remains a high-risk investment due to cash burn and no marketed product yet.
Dig Deeper ⛏️
P.S. Liking this issue? Forward to a friend 🧑🤝🧑
Headhunted 🦅
Recently funded private companies need talent! Scout jobs at recently funded European startups, ahead of your competition. 💪
Flower 🇸🇪 - The energy tech firm has closed a €20m Series A round. Developers, operations & finance roles open (link)
Vinted 🇱🇹 - The second-hand marketplace has raised €340m. Multiple roles available across Europe (link)
Paretos 🇩🇪 - The decision intelligence AI startup has raised an additional €8.5m in its Series A funding round. Sales & developer roles open. (link)
Modash 🇪🇪 - The influencer marketing platform has raised a $12m Series A. Multiple roles available across Europe (link)
Filigran 🇫🇷 - The cybersecurity firm has raised a $35m Series B round. Multiple roles open (link)
Interestingness📔
📚 New to investing? Grab a PDF copy of my ebook here.
As always, the financial disclaimer!
This is not investment advice. I am not a financial advisor. Make sure to conduct your thorough research before purchasing or selling financial products.