What are the best AI drug discovery software platforms? 

Author

  • Scott McDonald

Artificial Intelligence is playing a major and increasing role in drug discovery by accelerating the development of new medicines and reducing the associated costs. Traditional drug discovery is a time consuming and expensive process (many years and many billions of dollars), and carries significant risk, with many drug candidates failing in clinical trials.

AI-powered software can help overcome some of these challenges by, for example, analysing and providing insight into the sparse and noisy data available to drug discovery organisations, predicting drug-target interactions and designing new compounds that stand a greater chance of success.

What value does AI offer in drug discovery? 

The potential is huge: 

  • Faster drug development: AI can speed up early-stage research by identifying compounds that have a good chance of success in very short time frames (weeks rather than years). 
  • Cost savings: By focusing resources on compounds and assays likely to give the greatest return on investment, AI can lower R&D costs. 
  • Greater innovation: With a clear objective, AI can use the resources is has available to give insights with less bias than humans, for example to identify chemistries that no human would ever think to explore. 
  • Improved compound design: AI can generate new compound ideas that are optimised for your objectives. 

To learn more about the value we’re seeing from AI, you can read an article on AI transforming drug discovery chemistry by my colleague, Tamsin Mansley. 

What AI drug discovery platforms are available? 

There are many organisations developing AI platforms for drug discovery, and it would be impossible to cover all of them in this article. Here, I have chosen to highlight three platforms, providing a high-level overview of their capabilities and use cases. However, that does not mean other platforms are not suited to your needs. 

ProductSummaryWhat does it enable? 
DeepMirror Utilises predictive modelling and generative AI to streamline and advance molecular design – Generation and optimisation of new compounds 
– Acceleration of hit-to-lead workflow through in silico prioritisation 
– Learning from user data to refine molecular suggestions 
Cerella Leverages deep learning imputation to predict compound properties from sparse drug discovery data – Confident selection of the most promising compounds to advance 
– Identification of hidden opportunities by highlighting unlikely experimental values 
– Strategic prioritisation of experiments with the highest value 
Pharma.AI End-to-end drug discovery platform covering target identification to clinical trial design – Generation of new compound ideas via deep generative models 
– Guided target identification using multi-omics data integration 
– Acceleration of preclinical candidate selection and optimisation 

How to choose the right platform for you? 

There are a number of factors to consider here: 

  • Your objective: You first need a good understanding of your project aims. What are you trying to achieve? Is it the generation of new compound ideas, is it understanding risk in a project, is it highlighting existing compounds that may have been overlooked? Is it all of those or something else? 
  • Data availability: All modelling methods require at least some data, and AI-powered methodologies are no different. However, some methods are much more capable of dealing with sparse or limited data sets. What data do you have available? What is the quality and format of your available data? 
  • Computational need: Some platforms are computationally expensive, requiring significant amounts of time, processing power or memory. Do you have the in-house resources and expertise to run them? Have you considered a fully hosted solution? 
  • Integration: How are you going to integrate AI into your existing workflows? AI solutions provide the greatest return when they are fully embedded into the discovery process, and the output can be easily accessed and augmented with expert human knowledge. 

And in addition to this, you will need to be very clear on how you are going to measure whether your AI platform will deliver a return on investment. Often the objective and subsequent selection of a platform are the easiest steps to take, with little thought given as to how success (or failure) will be measured. If you’d like to learn more about evaluating the success of AI platforms, you can read an article on maximizing the ROI of AI by my colleague, Charlotte Wharrick. 

Conclusion 

AI presents a great opportunity for pharmaceutical and biotech organisations looking to enhance and streamline drug discovery R&D. While selecting the right platform requires careful consideration of your specific needs, resources, and technical capabilities, the potential rewards in terms of time savings, cost reduction, and innovation are substantial. 

As these technologies continue to evolve at pace, those that can successfully implement AI-powered approaches will likely gain a competitive advantage in bringing novel drugs to market. Success lies not just in adopting these powerful technologies, but in integrating them effectively with your infrastructure, and invaluable in-house expertise, to maximise their impact. 

About the author

Scott McDonald, MSc

Scott is a Business Development Manager at Optibrium, where he helps scientists make scientifically robust decisions through software solutions that enhance drug discovery and development. He works with clients to leverage Optibrium’s products.

With over 17 years of experience in the life sciences industry, Scott holds a Master’s degree in Chemistry from the University of Salford.

Linkedin

Scott McDonald

More AI in drug discovery resources