Pharmaceutical research and development is time-consuming and notoriously costly, often taking years and hundreds of millions of dollars. Yet, Insilico Medicine, a pioneering drug discovery company, is making strides to change this narrative through the implementation of generative AI in drug development.
Insilico, an esteemed member of NVIDIA Inception, is a testament to the potential of AI-driven drug discovery. Their success is best demonstrated in their breakthrough accomplishment of reaching Phase 2 clinical trials with a drug candidate for idiopathic pulmonary fibrosis. This severe respiratory disease, characterized by a progressive decrease in lung function, stands to be better tackled with this AI-discovered drug candidate.
Harnessing the power of generative AI, Insilico has streamlined the preclinical drug discovery process. From isolating a molecule for drug targeting to predicting clinical trial outcomes, AI plays a significant role. The beauty of this process lies in its efficiency. Traditional drug discovery methods could easily set a company back by over $400 million and span up to six years. Conversely, Insilico’s AI-enabled approach cost just a tenth of the price and took only one-third of the time.
“Translating biology and chemistry into deep learning is at the core of our approach,” stated Alex Zhavoronkov, CEO of Insilico Medicine. “Our first drug candidate entering Phase 2 marks a significant milestone, not just for us but for the entire AI-driven drug discovery landscape.”
As a premier member of NVIDIA Inception, a support program for advanced startups, Insilico utilizes NVIDIA Tensor Core GPUs in their AI drug design engine, Chemistry42. This allows them to generate novel molecular structures at an unprecedented speed. This early adoption of technology demonstrates the innovative spirit of the company.
Insilico’s Pharma.AI platform comprises various AI models trained on millions of data samples. These AI tools can rapidly identify and prioritize targets that have significant roles in a disease’s impact, such as the notorious spike protein in the COVID-19 virus.
Their generative chemistry tool, the Chemistry42 engine, uses deep learning to generate potential drug compounds that target the proteins identified by another AI tool, PandaOmics. Insilico’s approach applies deep learning to both biology and chemistry, a departure from the typical drug discovery AI companies that focus on either field.
To develop their pulmonary fibrosis drug candidate, Insilico used Pharma.AI to design and synthesize around 80 molecules, setting a new bar for the success rate of preclinical drug candidates. The complete process took less than 18 months.
As the drug undergoes Phase 2 clinical trials in the U.S. and China, Insilico is not resting on its laurels. The company has over 30 programs in the pipeline, targeting a range of diseases, including several types of cancer.
“Initially, many doubted that generative AI could achieve such diversity, novelty, and accuracy,” Zhavoronkov stated. “But with a pipeline of promising drug candidates, it’s clear that our method works.” This disruptive approach not only showcases the tangible benefits of AI in drug discovery but also paves the way for future innovations.
