HONG KONG – Insilico Medicine Hong Kong Ltd. has inked a two-program drug discovery collaboration agreement with Jiangsu Chia Tai Fenghai Pharmaceutical Co. Ltd. (CTFH), of Jiangsu, China. The deal will focus on drug discovery and development for previously undruggable oncology targets through Insilico's artificial intelligence (AI)-enabled platform.
"One of the main topics of this collaboration is indeed going to be triple-negative breast cancer as one of the toughest forms of the disease with unrevealed treatment targets," Alex Zhavoronkov, founder and CEO of Insilico, told BioWorld.
The collaboration could fetch the company up to $200 million for the achievement of milestone payments and royalties based on net sales for products from the collaboration.
"As the premier AI drug discovery company in the industry, Insilico Medicine has demonstrated the capabilities to generate novel molecules with specified properties using its next-generation AI platform," said Wenyu Xia, the general manager of CTFH.
Xia stated that this collaboration would "speed up the R&D process, reduce the cost and provide greater benefits to patients."
The speed in which Insilico's AI platform works is startling indeed. Just last month, Hong Kong-based Insilico published a landmark paper in Nature Biotechnology that demonstrated the application of its generative tensorial reinforcement learning (GENTRL) systems in the generation of novel molecules for simple kinases.
It generated 30,000 designs for molecules that target a protein linked with fibrosis in only 21 days. Six of those molecules were synthesized in the lab, two of which were tested in cells, with the most promising one tested in mice. The eventual conclusion from the experimental validation was that it was potent against the protein and showed drug-like qualities.
The entire process took only 46 days.
"The CTFH project is aimed at first-in-class medicines, and we need to use [a] much more sophisticated AI pipeline; therefore [the] exact timeline is being defined yet," said Zhavoronkov. "But we are expecting to see the first results in around a year's time and may be able to talk about it in just over two years."
Insilico's system differs from the standard AI drug development, which involves screening millions of potential molecular structures looking for a viable fit, by focusing instead on an algorithm that imagines potential protein structures based on existing research and certain pre-programmed design criteria.
"The drug discovery platform starts with target identification and rapid target validation using generative chemistry. We also use the clinical trials outcomes prediction engine to improve the probability of successfully passing clinical trials right at the target selection and molecular generation stage," said Zhavoronkov.
That concept of Generative Adversarial Networks (GANs) is relatively new and is often referred to as the "AI imagination," "creative AI" or "AI curiosity."
Conceptually, Insilico describes it is a competition between two deep neural networks, where one, the generator, is generating novel content with the desired set of criteria. And another, the discriminator, is testing whether the output of the generator is true or false.
"In 2016, multiple groups using GANs started producing new photorealistic images from natural language. For example, one could give a description: 'This small bird has a pink breast and crown, and black primaries and secondaries,' and the GAN would generate or 'imagine' a large number of images of birds with said properties," said Zhavoronkov.
Insilico started working on an application to incorporate GANs and the generation of novel chemical structures or molecules in 2015.
"When generating pictures, GANs require high-dimensional data and large, well-annotated training sets. Molecules can be represented in the low-dimensional format like binary fingerprints, SMILES strings, graphs and other light representations that can be used to synthesize the resulting molecules," said Zhavoronkov.
Not surprisingly, Insilico won the attention of those in both the biotech and AI sectors with the publication of that paper. Shortly after that, it announced a $37 million round led by prominent biotechnology and AI investors.
"The series B funding will be used to commercialize the validated generative chemistry and target identification technology," said Zhavoronkov.
The company will also build up a senior management team with the experience in the pharmaceutical industry, further develop its pipeline in cancer, fibrosis, nonalcoholic steatohepatitis, immunology and CNS diseases for the purposes of partnering with the pharmaceutical companies on specific therapeutic programs.
Insilico claims it has a long list of collaborations, many of which are with startup companies that are willing to incorporate the AI component into their discovery-stage efforts and also generate the data in more sophisticated ways so that the data could be re-used in the future.
"The CTFH collaboration is the biggest in China so far. But we have several other joint works running or just about to get started," said Zhavoronkov.
"Among them are, for example, collaborations with chemical industry partners to get huge molecule databases run through by AI for effective search of potential drugs."