Machine learning leads to breakthrough in drug discovery

An advanced machine learning algorithm developed by an international team, including Dr James Kermode of the University of Warwick, could drastically speed up the discovery of new drugs. Researchers from seven different academic institutions across the world have developed the highly-efficient machine learning algorithm – a potential breakthrough in drug discovery.

Dr James Kermode, from the Warwick School of Engineering, helped devise an algorithm which can predict whether a candidate drug molecule will bind to a target protein. The algorithm can do this with 99% accuracy, given only a relatively small set of training data.

The algorithm can do this with 99% accuracy, given only a relatively small set of training data

The algorithm is said to capture the quantum mechanical effects governing the complex surface reconstructions of silicon, predict the stability of different classes of molecules with chemical accuracy, and even distinguish between active and inactive protein ligands. Dr Kermode explains that this research provides a general-purpose machine learning approach which can be applied to both materials and molecules.

“The research is expected to lead to a significant increase in the accuracy and transferability of models used for drug design and to describe the mechanical properties of materials,” adds Dr Kermode. The new method will allow scientists to predict the behaviour of new materials and molecules with great accuracy and little computational effort, saving time and money.

The new method will allow scientists to predict the behaviour of new materials and molecules

The algorithm combines local information from the surrounding area of each atom in a structure, which makes it applicable across different classes of chemical, materials-science, and biochemical problems. This new approach has also shown remarkable success in predicting the stability of organic molecules.

This research reveals how chemical and materials discovery is benefitting from machine learning and artificial intelligence (AI). Various prominent institutions recognise the potential of computer science in many areas of medicine, including drug discovery, diagnosis, and epidemic outbreak prediction. Google has a division of their DeepMind project devoted entirely to healthcare. DeepMind began working with the NHS to conduct research into the applications of their AI in hospitals. An early app by the company, Streams, uses a range of data to identify patients at risk of acute kidney failure.

This research reveals how chemical and materials discovery is benefitting from machine learning and artificial intelligence

This research has led to a breakthrough in drug discovery, given that it could accelerate the screening of candidate drug molecules thousands of times over. Thus, this new research has the potential to revolutionise healthcare, so the future of machine learning looks extremely bright.