Written by:
Senior Director, Biologics Engineering, Oncology R&D, AstraZeneca
Executive Director, Head of the Centre for Artificial Intelligence, BioPharmaceuticals R&D, AstraZeneca
As part of our commitment to innovation, we are harnessing the power of artificial intelligence (AI) to transform drug discovery and design. In pharmaceutical research and development (R&D), we have shifted from traditional wet lab processes to integrated computational methods which augment established workflows.
Deeper integration of AI in drug discovery
Artificial intelligence has become an essential and integrated part of the science we deliver at AstraZeneca. It's no longer just an 'add-on' or a nice-to-have; AI now gives us speed, scale, and insights that were previously not possible, accelerating and enhancing the creation of potentially life-changing medicines.
Groundbreaking research collaborations
Our recent collaborations with the University of Sheffield and the University of Cambridge have led to two publications that showcase the depth of our AI integration and its potential to revolutionise drug discovery. These breakthrough technologies demonstrate how AI approaches are transforming the drug development process for discovery and design, and across therapeutic modalities.
MapDiff: Making it easier to design protein-based drugs to treat diseases
In collaboration with the University of Sheffield, we've developed MapDiff, an innovative AI framework for inverse protein folding - a critical protein engineering process. This research, published in Nature Machine Intelligence, outperforms existing methods and represents a significant leap forward in our ability to design novel therapeutic proteins with specific functions.
Understanding inverse protein folding
For medicines to work properly, proteins need to fold into very specific 3D shapes. This new approach works like a guide that predicts the most important folds in the protein structure, making the design process faster and more accurate. While traditional protein folding aims to predict a protein's 3D structure from the amino acid sequence, the building blocks of proteins, inverse protein folding does the opposite. This is like designing a specific origami shape by unfolding an existing shape, seeing the lines on the paper and using these to determine the correct sequence of folds for the next origami.
Edge Set Attention: Enhancing molecular property prediction to identify promising drug candidates
Our collaboration with the University of Cambridge has led to the development of a new AI model, Edge Set Attention (ESA) for predicting key molecular properties of a potential medicine using what’s called a graph attention approach. This research, published in Nature Communications, significantly outperforms existing methods and improves the ability to predict how potential drug molecules will behave, helping us identify promising drug candidates.
Unlocking drug discovery with graph-based AI models
Graph based AI models are particularly well-suited for analysing molecular structures. They represent molecules as graphs, where atoms are nodes and chemical bonds are edges. This allows the AI to learn and predict molecular properties based on the structure and connectivity of the molecule - essential for understanding drug efficacy and drug safety profiles.
Transforming drug development with AI approaches
AI innovations are fundamentally changing how we approach pharmaceutical research. By integrating advanced technologies, such as machine learning models and deep learning technologies throughout our early drug discovery pipeline, we're able to:
- Accelerate drug discovery: More than 90% of our small molecule discovery pipeline is AI assisted, and we are rapidly integrating AI and machine learning approaches into discovery and design of next generation medicines such as biologics and peptides, to increase our probability of clinical success.
- Advance early detection and disease insight: By harnessing AI we can spot patterns and connections with potential to help design more effective and tolerable medicines, develop optimal medicine delivery methods, and monitor disease progression enabling earlier intervention.
- Transform clinical trials: AI tools are helping us to design efficient and smarter clinical trials by selecting patients who are most likely to respond to treatment and by measuring endpoints that more accurately predict the potential benefits of new medicines for patients – helping us get the right medicine, to the right patient, at the right time.
The future of AI in drug discovery
The two AI approaches outperform existing methods, exemplifying how we at AstraZeneca leverage AI to push the boundaries of what's possible in drug discovery and design. By combining our expertise in data science, biology, and chemistry, we're accelerating the path from concept to potential new therapies.
As we continue to integrate AI across every aspect of R&D, we're not just improving existing processes — we’re reimagining drug discovery. Our AI approaches span from initial biological data analysis through to transforming clinical trials, always with the goal of delivering potentially life-changing medicines to patients faster and more efficiently than ever before.
The pharmaceutical industry is experiencing a fundamental transformation through artificial intelligence. Machine learning methods, neural networks, and advanced computational approaches are becoming integral to how we understand human health and develop therapeutic solutions.
Looking Ahead: We continue to explore emerging AI technologies, including quantum computing, foundational models and agentic AI, to further enhance our drug discovery capabilities. Our partnerships with academic institutions and AI companies worldwide strengthen our ability to stay at the forefront of this technological revolution.
Explore opportunities with AstraZeneca today
Ready to learn more about AI and data science at AstraZeneca across therapeutic areas? Discover how we're transforming drug discovery at scale.
