In the early stages of finding new drugs hundred thousands of molecules are tested for activity. This so-called high-throughput screening is performed using costly robotic equipment. Often it leads to wrong results, which could be false positives or false negatives. Computer-based virtual screening can reduce the high-costs and time required. However, it provides the same or even worse accuracy compared to lab-based screening.
Dr Andreas Kukol, a senior lecturer in Biochemistry at the University has developed methods that combine the results from different virtual screening algorithms in a consensus approach. The results show that some consensus methods can achieve a higher accuracy in predicting active drugs than individual methods on its own. Dr Kukol said that more efficient computational tools can revolutionise the drug discovery process and shorten the 15 years currently required to develop a new drug.