Today’s drug discovery research has seen a paradigm shift since the advent of computational biology. The tools of computational biology have made it easier to search for possible drug molecules for particular diseases. It has become feasible to screen possible drug molecules and see how those molecules would work in a cellular environment. To that end, the drug industry today is collaborating with computational biologists, precisely those engaged in bioinformatics.
In a recent study published in Nature, researchers scanned a vast chemical database of about 170 million molecules, which is almost 100 times larger than previous databases, and identified new compounds that have the potential to be starting points for novel antibiotics and antipsychotic medicines. It is expected that the resource would grow up to more than 1 billion molecules in the next year, making the technique powerful with time.
Brian K. Shoichet of University of California, San Francisco, USA, and his colleagues screened 170 million molecules that are stored in the database of Enamine — a chemical supply company. The screening was aimed at finding drug compounds against two targets — AmpC β-lactamase, a focal point of some antibiotics, and the D4 dopamine receptor protein, a target of potential antipsychotic medications. But there was a challenge that the researchers had to overcome — the sheer size of the database. “We were terrified as to how we were going to find the interesting molecules, spotting the signal in the noise,” Shoichet told Science.
The researchers adopted an approach where they first tested whether their computational tools can find the hundreds of already known inhibitors of the two targets, that they were searching drugs against, amongst the 170 million molecules in the database. They used a parallel computing approach where they used a cluster of 2,000 computer processors. They found that the top scoring molecules were those already known as inhibitors and their structural cousins. Next, Enamine scientists made hundreds of previously-unidentified compounds that were high-scoring. Out of these compounds, 24% were found to bind with the D4 receptor with high affinity and 11% against AmpC β-lactamase. This was a far higher hit rate in comparison to other virtual screening programs.
Shoichet had previously done virtual screening in drug designing in 2016, where his team screened for 3 million molecules of possible drugs. These 3 million molecules were also taken from the Enamine database. The team found a potential opioid painkiller that would lack the addictive properties of the currently used opioid drugs. Epiodyne, a biotech company, is now working to turn the screened molecule into a drug.
About Virtual Screening:
Researchers use virtual screening to scan possible drug targets. In this approach, researchers computationally evaluate how well a potential molecule might bind to a protein or any other biological target in the body. Generally, scientists use the software called ‘Molecular Docking’ to analyze the thousands of possible orientations that a molecule can take upon binding with its molecular target. Out of these, the orientations that bind most tightly are identified for experimental testing.
But the real problem lies in the extraordinarily large number of possible drug-like molecules. The number could be 1063, which is similar to the number of atoms in the universe. To ease the problem, researchers team up with chemical supply companies that can make vast libraries of molecules. One such company is Enamine. It started with some 70,000 chemical building blocks that can interact with one another through 130 well known chemical reactions. This enabled the company to build a database of 700 million compounds.