Insilico Medicine, a company specializing in AI-driven drug discovery, recently announced a successful collaboration between quantum computing and generative AI in the field of lead candidate discovery for drug development. The study, published in the Journal of Chemical Information and Modeling, was led by Insilico’s Taiwan and UAE centers, which focus on pioneering breakthrough methods using generative AI and quantum computing to accelerate drug discovery.
The research received support from Alán Aspuru-Guzik, Ph.D., director of the University of Toronto Acceleration Consortium, and scientists from the Hon Hai (Foxconn) Research Institute. The international collaboration aims to advance AI in drug discovery and has brought together experts from various organizations, including Foxconn, Insilico, Zapata Computing, and the University of Toronto.
Generative Adversarial Networks (GANs) have proven to be successful generative models in drug discovery and design. These models consist of a generator and a discriminator, where the generator creates data that mimics a specific distribution, and the discriminator distinguishes between real and fake samples. The GAN is trained until the discriminator can no longer distinguish between the generated and real data.
The published paper explores the application of quantum computing in small molecule drug discovery by substituting components of MolGAN, an implicit GAN for small molecular graphs, with a variational quantum circuit (VQC). The researchers compared the performance of the quantum GAN with its classical counterpart, examining the noise generator, generator with the patch method, and quantum discriminator.
The study demonstrated that the trained quantum GANs, using the VQC as the noise generator, can generate molecules similar to those in the training set. Moreover, the quantum generator outperformed the classical GAN in terms of drug properties of the generated compounds and goal-directed benchmarking.
The researchers also showed that the quantum discriminator, with only a few learnable parameters, can generate valid molecules and outperforms the classical counterpart with tens of thousands of parameters in terms of generated molecule properties and KL-divergence score.
Insilico Medicine plans to integrate the hybrid quantum GAN model into Chemistry42, the company’s proprietary small molecule generation engine, to further enhance their AI-driven drug discovery and development process.
Insilico has been at the forefront of using GANs in de novo molecular design and published the first paper in this field in 2016. The company has successfully generated 11 preclinical candidates using GAN-based generative AI models, and their lead program has been validated in Phase I clinical trials.
Alex Zhavoronkov, Ph.D., founder and CEO of Insilico Medicine, expressed pride in the achievements of the quantum computing team and their innovative work. He sees this as the first step in their journey and mentions ongoing experiments with real quantum computers for chemistry. Insilico aims to share its best practices with the industry and academia, contributing to advancements in the field.
Source: Insilico Medicine
