Etienne Bonanno, an M.Sc. student in Artificial Intelligence, together with his dissertation supervisor, Dr Jean Paul Ebejer from the Centre for Molecular Medicine and Biobanking, have recently published the findings resulting from Etienne's dissertation in a paper titled '' in prestigious, peer-reviewed journal Frontiers in Pharmacology.
Ultrafast Shape Recognition (USR) is a computational technique in Cheminformatics that is used to search through massive electronic databases of chemical compounds to find molecules that are structurally similar to one or more target molecules which are known a-priori to exhibit drug-like properties. According to the Similarity Principle Principle, molecules that are similar to each other will have similar properties, therefore new molecules that are similar in shape to existing targets have a high probability of having similar drug-like properties by chemically binding to the same proteins. This class of techniques is known as Ligand-Based Virtual Screening.
In this research, the authors augmented the standard USR method by applying a selection of Machine Learning techniques in order to improve the recognition rate of the algorithm, in effect recognising correctly more molecules that will likely have the desired drug-like effect. By applying machine learning to standard USR, improvements ranging from 100% to 900% in recognition rate were obtained in studies based on a standard electronic molecular compound database, the Database of Useful Decoys - Enhanced (DUD-E).
The study was published in , an open-access online journal which is a leader in its field, publishing rigorously peer-reviewed research across disciplines, including basic and clinical pharmacology, medicinal chemistry, pharmacy and toxicology.
For further information about this research contact Dr Jean Paul Ebejer.
