As the world is three-dimensional, AI systems need the ability to learn about three-dimensional spatial relationships. Often the best way to attack an abstract problem like representing spatial information, is to first tackle a specific concrete problem, and then to generalise to the abstract. This project therefore focuses on the specific problem of representing the three-dimensional binding of drugs to drug-targets.
Explaining the science
One approach to spatial learning is the development of probabilistic relational learning methods for learning three-dimensional spatial relationships. This approach differs from standard machine learning in that it uses first-order predicate logic as the basic representational language, in contrast to propositional logic, which is the basis of machine learning methods such as neural networks. A proposition is a statement that is true/false about a whole object, which makes the representation of 3D spatial relationships difficult without a standard reference frame. Relational learning is well suited for representing chemical structure, as the standard human-chemist representation of chemical structure is as 3D relationships between atoms and bonds.
For information about rough path theory see another Turing project that is using it for streamed data here.
The project proposes to compare and contrast two complimentary approaches to representing 3D space. The first is to apply relational learning and rough path theory to the specific problem of representing the binding of drugs to protein active sites. The second is to integrate relational learning and rough path machine learning for structure based drug design.
The researchers will compare the predictive success of the integrated method with the individual methods, and other state-of-the-art methods.
The application of machine learning to drug design is central to the success of the UK pharmaceutical industry, and its ability to develop novel drugs. Discovering an developing new drugs is generally slow (often taking more than 10 years) and expensive (costing more than $1 Billion).