Is the Market Underestimating the Potential of Exscientia?

Exscientia plc is one of several companies harnessing the power of artificial intelligence (AI) to address the low returns on drug discovery investments. While AI has the potential to create enormous value, there are still questions about its overall impact on R&D efficiency, its ability to develop truly novel therapies, and the sustainability of any advantages it provides.

Unlike some of its peers, such as Recursion Pharmaceuticals, Inc. (RXRX), Exscientia hasn’t yet fully capitalized on AI hype. However, this could present an opportunity for strong returns for shareholders if the company’s approach starts to yield results or gains more investor attention.

The Challenge of Drug Discovery

Drug discovery typically involves a trial-and-error process to identify a molecule that safely induces the desired biological effect in humans. Given the complexity of the human body and the vast molecular search space, this is a slow and costly process with high failure rates.

For instance, there are an estimated 10^60 possible molecular combinations with drug-like characteristics, making it impractical to explore even a tiny fraction of this space. A typical drug discovery project can only afford to synthesize and test fewer than 10,000 molecules.

Over the past few decades, drug discovery has faced increasing challenges. New drugs have a 96% failure rate from project inception to approval, and it costs roughly $1.8 billion and over 10 years to bring a new drug to market.

As a result, the projected return on investment (ROI) in R&D for the top 12 global pharmaceutical companies has fallen to 1.8%, with the risk-adjusted net present value (NPV) of an asset at around $10 million. This situation has contributed to only a trickle of new drugs reaching the market each year, with the FDA approving 42 drugs in 2019.

Addressing Key Issues in Drug Discovery

Exscientia believes that several key issues must be addressed to improve drug discovery outcomes:

• Drug design is a complex multi-dimensional optimization problem.
• Translational models do not represent the complexity of human disease.
• Dose-limiting effects can limit clinical efficacy.

With advances in laboratory tools and AI, it is now feasible to significantly improve drug R&D outcomes. Techniques are available to reduce the difficulty of synthesizing and screening new chemicals, and computational tools aid the design and testing of new drugs.

Exscientia’s AI-Driven Approach

Exscientia initially focused on small molecules, although it is now expanding its platform to include other types of therapies. Small molecules represented the majority of approved therapeutic drugs in 2019 and accounted for 75% of the $1.2 trillion in drug sales.

Exscientia chose this focus because small molecules can perform biological functions such as intracellular activation or inhibition that are not possible with other modalities and can be easily distributed into the brain.

AI Applications in Drug Discovery

AI can be applied in various areas of drug discovery and development, including:

• Improving understanding of disease biology.
• Identifying drug targets.
• Designing molecules.
• Optimizing molecules.

Exscientia takes a comprehensive approach to AI-driven drug discovery, using AI to identify targets, design drug candidates, and select patients. AI is also utilized in translational models to enable efficient exploration of chemical space, generating novel molecules with improved properties.

However, AI-generated compounds can be difficult to manufacture and may contain reactive groups that cause safety issues. Additionally, AI’s reliance on training data raises questions about its ability to extrapolate beyond known data and generate truly novel molecules.

Exscientia’s Process and Platform

Exscientia’s drug discovery process involves several key components:

1. Precision Target: AI is used to prioritize projects.
2. Precision Design: AI is used to design drugs.
3. Precision Experiment: Tech-enabled experimentation generates better data.
4. Precision Medicine: Integrated analysis of patient data ensures clinical relevance.

Exscientia’s platform leverages a range of structural and biochemical data and various algorithms, including evolutionary, reinforcement learning, deep learning, and active learning. AI design is integrated with high-throughput screening to train models and guide active learning.

Exscientia also uses its platform for biomarker discovery and patient stratification, setting it apart from many other companies in the space. The company believes its approach offers several advantages:

• Patient Relevance: Using patient cells may help pre-clinical research translate into clinical success and assist in patient selection.
• Single-Cell Resolution: Can help distinguish on-target from off-target responses, identifying effective and non-toxic drugs.
• Versatility: Compatible with a broad range of cancers and tissue types.
• Scalability: Automation and high-throughput lab equipment enable the analysis of many drugs in parallel.
• Speed: Turnaround times per assay can be as low as five days.
• Reproducibility: Provides highly reproducible results within and between assay runs.

Clinical Studies and Programs

Exscientia has demonstrated the ability to create novel optimized drug candidates faster than the industry average. The company’s process averages one year from AI generation of a first novel molecule to the design of a development candidate, compared to the industry average of around 4.5 years.

The company also synthesizes fewer compounds compared to conventional approaches and advances around 30 programs concurrently, despite limited resources.

Target Selection and Drug Design

Selecting the best molecular or cellular target is crucial in drug discovery. Exscientia uses its Centaur Biologist target analysis platform, combining literature analysis with other data in a knowledge graph to generate novel gene-disease target hypotheses.

Drugability is assessed using Exscientia’s protein mapping technologies to prioritize targets based on tractability, and experimental data is used to test hypotheses and ensure target effectiveness.

For drug design, Exscientia uses AI to anticipate the characteristics a drug will need, optimizing drug candidates across multiple design objectives. The Centaur Chemist platform combines AI with models to generate synthetically tractable and drug-like molecules with desirable properties.

Data Acquisition and Clinical Studies

Exscientia’s drug discovery approach relies on access to high-quality and high-dimensionality data. The company has invested in laboratory capabilities to generate screening data and structural biology data. Exscientia uses various technologies to ensure lab results translate to in vivo success, leveraging its understanding of disease biology and population heterogeneity.

In the EXALT-1 clinical study, Exscientia’s platform predicted the effectiveness of potential cancer treatments using AI and patient samples, achieving a 55% overall response rate and a statistically significant improvement in progression-free survival over prior therapies.

However, Exscientia does not currently plan to use AI to treat patients in a clinical setting.

Valuation and Risks

Exscientia believes its platform increases the NPV of projects by 4x compared to the industry average, driven by improved success rates and time savings. Despite this, the company faces several risks. Early data from AI-enabled drug discovery efforts is promising but insufficient to determine the overall impact on drug development.

Questions remain about AI’s ability to develop truly novel drugs and its role in navigating existing patents.

Conclusion

Exscientia is taking an integrated approach to AI-enabled drug discovery, leveraging diversity of biology in both drug discovery and patient selection. While the company lacks the scale of competitors like Recursion, it has a relatively small enterprise value, presenting potential for significant returns if its approach yields positive results.

Lower interest rates or positive clinical data could see the company’s stock move significantly higher, despite current negative sentiment.