Scientists have created a system called AdaptiveFlow that makes it possible to virtually sift through billions of potential drug molecules at about 1,000 times lower computational cost than previous methods. In principle, this could shorten the initial drug-discovery stage from months to hours — and significantly expand the chemical space available to researchers.
Why does this matter? Chemical space is staggeringly large. The more of it can be screened in practice, the greater the chance of finding a promising molecule. AdaptiveFlow is being released as open source, meaning its code is freely available for other researchers to use and develop further.
The system selects the most promising compounds and then checks their ability to bind to target proteins. It has already been tested on two cancer targets and found strong candidates for both. Computer-based drug discovery is gaining a new scale: far more molecules can now be screened before the work reaches the laboratory.
