Alphabet, Google’s Parent company, launches an AI-driven Drug Discovery Startup

Artificial Intelligence (AI) has had a major impact on the field of drug discovery, transforming a large practical problem into a manageable information problem. Alphabet, the company that owns Google, is the most recent to do so, forming Isomorphic Labs under DeepMind CEO Demis Hassabis to pursue this exciting new topic.

11 Startups to Participate in First Africa Artificial Intelligence Accelerator Cohort

Few details about Isomorphic Labs were given in it’s first blog post and accompanying FAQ. The company’s mission statement reads as follows: “to build a computational platform to understand biological systems from first principles to discover new ways to treat disease.” 

This assertion is based on a few assumptions, most notably that it is possible to simulate biological systems computationally in a way that facilitates drug development.

In recent years, some massive startups have been established and funded with hundreds of millions of dollars to pursue very similar goals, but there has been no noticeable revolutions or spectacular AI-discovered remarkable drug for a hitherto incurable disease. But it’s evident that these AI systems aren’t miracle factories, just elements in a long and intricate process that still needs a lot of time, money, and test tubes.

AI has long been touted as a tool to speed up and lower the cost of discovering new medicines to treat a wide range of ailments. AI could be used to help researchers locate molecules that best fit a specific biological goal, for example, or to fine-tune the proposed compounds. Over the past two years, hundreds of millions of dollars have been invested in companies developing AI capabilities. 

It may be more difficult to develop and test medications than it is to figure out the structure of proteins. There is no way to predict how strongly two proteins will adhere to each other, even if their shapes are physically compatible. When administered to an animal or a person, a medication candidate that appears promising based on how it works at a chemical level may or may not work at all. 

Scientist and Writer Derek Lowe published a paper in Science this summer stating that over 90 percent of medications that make it to a clinical trial fail. There isn’t anything wrong with the molecular level in the majority of cases that produce issues.

Isomorphic’s suggested work at DeepMind could help break through some research bottlenecks, but it isn’t a simple panacea for the myriad obstacles in drug development. This type of time-consuming, resource-draining work will continue to be done by structural biologists like Helen Walden at universities around the country.