AI-Driven Drug Development: The Future of Virtual Screening of Lead Compounds

Drug Discovery Entering a New Era

Researchers from University of Washington have identified an AI-accelerated virtual screening that can enable the very fast and precise identifications of drug candidates from ultra-large chemical libraries, thereby changing the paradigm of drug discovery. 

Drug discovery has always been painful and burdening. Traditional in silico and in vitro approaches have relied heavily on physical screenings of large chemical libraries. This represents considerable resources. In any case, all that is starting to change with the integration of AI into virtual screening platforms. A recent study, “An Artificial Intelligence Accelerated Virtual Screening Platform for Drug Discovery,” proposed by Guangfeng Zhou et al., is an up-and-coming innovative approach that uses the power of AI in accelerating virtual screening for drug development.

The Power of AI in Virtual Screening

Virtual screening represents an important component in the workflow of drug discovery for the early identification of promising compounds. Complete screening is becoming prohibitively expensive and time-consuming as chemical libraries grow to billions of compounds. In this study, the authors present RosettaVS, a very accurate structure-based virtual screening method that predicts with a high level of accuracy both the docking pose and the binding affinity of a given compound by taking into consideration the flexibility of the receptor in its modeling process; it outperforms state-of-the-art methods.

RosettaVS uses artificial intelligence to handle ultra-large libraries with ease by incorporating deep learning to accelerate the screening process. This means that one will be able to screen a mega-sized chemical library in seven days, a feat that would take months using conventional methods​

What Makes RosettaVs Standout?

​​One of the salient features of RosettaVS includes screening multi-billion compound libraries against specific drug targets. The researchers indeed had tested this AI-accelerated platform on two targets: a ubiquitin ligase target KLHDC2 and human voltage-gated sodium channel NaV1.7.

These screenings led to the identification, by RosettaVS, of seven hit compounds for KLHDC2 at a 14% hit rate and four hit compounds for NaV1.7 at a 44% hit rate. Binding affinities were in the single-digit micromolar range, showing efficiency of the platform in identifying promising lead compounds.

This AI-accelerated platform is more than a speedier way of screening; it overcomes many of the substantial weaknesses of more traditional physics-based methods of docking, due to their high cost and discouraging time consumption. By integrating deep learning with an open-source virtual screening platform-OpenVS-this work heralds the possibly widespread use of such a platform across both pharmaceutical and biotechnology industries. The platform in itself can easily scale or be customized for different targets and thus assumes a wide range of versatility for researchers around the world.

A platform that will revolutionize the industry.

RosettaVS and OpenVS are two AI-accelerated virtual screening platforms that take drug discovery to a completely new dimension. Surmounting the limitations of traditional methodologies, these tools grant unparalleled speed, precision, and flexibility in the identification of viable drug candidates. As pharmaceutical and biotechnology companies continue to embrace such advanced tools, we will see not only faster drug development but also more targeted and effective treatments. The future of drug discovery looks to be dependent on AI, and such a novel approach allows new standards in how far we can leverage technology to match the medical challenges of tomorrow.

Know more about this article here.