AI-driven drug discovery (2023)

1. Better drugs, faster?

   The pharma industry has been slow to welcome digitalisation but is in command of giant cheque books, so more bankrolling of agile early-stage startups may help turbo-charge more efficient drug development. Despite the long approval cycles and significant regulatory hurdles, many of the early-stage companies highlighted in our previous briefing are already showing traction. Parisian rising star Aqemia raised a €30m Series A in October 2022 to develop new AI that can massively increase the speed at which promising drug candidates are found using quantum physics, and has already partnered with incumbents Sanofi, Servier and Janssen.

2. Repurpose, reuse, recycle

   While its success at discovering new drugs has been limited, AI has proven highly effective in analysing the secondary use of approved drugs to repurpose them for conditions they weren’t originally designed for. For example, Clemastine, an antihistamine, has been shown to reduce repetitive behaviours, common in those suffering from autism.Without the need to start from scratch, the drug development process can move much faster, leading to more effective treatments for a wider range of illnesses while saving on R&D costs.

3. The protein-folding problem

   Machine and deep learning are increasingly being deployed to predict the structure of unknown proteins, determined by the sequence of 20 amino acids assembled by DNA. DeepMind (acquired by Google in 2014 for £400m) expanded the number of protein structures in its open-source database from 350k to 200m+ in July 2022. Widening access to a growing number of proteins will help accelerate others’ progress in accurately determining protein structures from just their amino acid sequencing — without the need for tedious and costly lab analysis.

4. Big health data

   The more data AI has to work with, the better the insights. That means massive amounts are needed to train AI models in protein design, with precision medicine being the ultimate goal. But Europe is suffering compared to the US: it’s just not easy for scientists on the continent to access that scale of data. At present it necessitates extensive collaborations between labs to process massive datasets, which includes health data from diverse sources such as clinical research, electronic medical records and smart wearable devices. Companies that possess advanced data generation capabilities will have the upper hand.

Who early stage startups are up against