New Search Engine Software Enables Scientists to Identify Molecules by Shapes,
Speeding Drug Lead Screening by up to 14,000 Times

Click HERE to View the Abstract

Click HERE to View the Abstract

Virtual screening is a key technique in computational drug discovery, aimed at identifying those drug-like molecules that are likely to have beneficial biological properties. It is an obvious way to reduce expensive biological tests and tackle the high failure rate currently faced by the pharmaceutical industry.

In molecular docking, for instance, the process of docking the screened molecules to a biological target (almost always a protein) is simulated to provide an estimate of the drug-like properties, such as the binding energy, of the molecule. These techniques have spurred the generation of massive databases of drug-like molecules. This is the case in our widely publicized Screensaver-Lifesaver project  in which a database of 3.5 billion compounds was screened against a series of protein targets over a grid of 3.5 million personal computers, each running a screensaver incorporating the binding energy estimation (For information about the Screensaver-Lifesaver Project, click HERE).

Alternatively, molecular databases are routinely screened for compounds that most closely resemble a drug molecule of known biological activity to provide novel drug leads with similar bioactive effect.

It is widely believed that three-dimensional molecular shape provides the most discriminating pattern of a molecule and is an essential indicator for its biological activity, as it is directly related to the steep repulsive part of the interaction potential between the drug-like molecule and its target. However, efficient comparison of molecular shape remains a challenge. Currently available methods require the alignment and superimposing of the molecules for shape comparison, which often leads to suboptimal results and increase the error rates. More importantly, as shape comparison is a complex process, current methods are simply too slow to screen the largest databases.

At the NFCR Center for Computational Drug Discovery at the University of Oxford, Dr. Pedro Ballester, together with Center Director Dr. Graham Richards, developed a new technique known as the Ultrafast Shape Recognition (USR) for molecular shape comparison. This new method has successfully circumvented the problems currently faced in this field. Researchers in the center show that this new approach based on analysis of the positions of the atoms within a molecule is able to recognize molecular shape at least 1,546 times faster than current methodologies. Such an ultrafast method permits the identification of similarly shaped compounds within the largest molecular databases in a short time. It is estimated that for a molecular database screening which takes at least 1.1 years to conduct using the best existing method, the same job can be done within 7 hours when using USR! In addition, the problematic requirement of aligning molecules for comparison is circumvented, as the proposed distributions are independent of molecular orientation.

This methodology could be also adapted to tackle similar hard problems in other fields, such as designing content-based Internet search engines for three-dimensional geometrical objects or performing fast similarity comparisons between proteins. From a broader perspective, it is anticipated that USR will soon become not only useful, but also essential, to address the challenge of data explosion currently experienced in most scientific disciplines.

For detailed description on the newly developed USR, click HERE.

For more information on the Screensaver-Lifesaver Project, click HERE.