Seven Companies Using AI to Create Drugs

Drug development is a complex business. Medicinal chemists, many of whom have spent nearly 10 years in training, must painstakingly synthesize every molecule that could potentially have a therapeutic effect. These compounds then need to be tested on cells to make sure they have a statistically significant effect on a model of a diseased cell. 

Then the "drug candidates" are tested on animals to get at least a little closer to the serious step — clinical trials. Only one in 1,000 potential drugs ever goes through human testing. The whole process takes an average of 12 years — during which experimental drugs begin their journey from the lab and end up in your home medicine cabinet. It involves up to $2 billion in costs and an endless number of failures.

Artificial intelligence and machine learning promise to put an end to this inefficiency. Investment in using AI for drug development received the largest share of all private AI investment in 2020 — more than $13.8 billion. Drug development is already an expensive enterprise, and everyone in the pharmaceutical industry needs all the help they can get. 

Let's look at seven companies that use AI capabilities to find a cure for cancer and other serious human diseases. 

Roivant Sciences

Founded in 2014, the New York-based company has attracted huge funding of $1.9 billion. It accelerates the development of late-stage "drug candidates" by building subsidiaries (which always end in "vant"). Today, Roivant's subsidiaries are developing more than 40 drugs. 

In February 2021, the company acquired Silicon Therapeutics to expand its computing capabilities for drug development. Silicon promotes a drug design and development approach based on physical principles. The strategy boils down to using first principles from physics to determine whether a "drug candidate" exhibits any biological activity. 

Insitro

Founded in 2018, the drug discovery and development company has attracted investment from Andreessen Horowitz and the Investment Board of the Canada Pension Plan. Insitro was founded by Dr. Daphne Koller, a professor of computer science at Stanford University. Dr. Koller and her team are tackling one of the problems that make it difficult to bridge computational biology and machine learning — data. 

Existing biological data is limited or of too low quality for machine reading. The company is building a pipeline of experimental data collected from stem cell research to create a huge array of datasets based on real biological systems. Using this data bank, Insitro will apply machine learning to predict how healthy and diseased cells behave inside the human body, and what happens when potential drugs are introduced into a simulation model of a diseased system.

XtalPi

The Cambridge-based company was founded by quantum physicists from the Massachusetts Institute of Technology in 2014 and has since attracted more than $780 million in investment from companies such as Sequoia Capital China and Google. 

XtalPi created the Intelligence Digital Drug Discovery and Development machine learning platform, which predicts the crystal structure and properties of small-molecule "drug candidates" using the principles of quantum mechanics and solid-state crystallography. In 2018, the company began collaborating with Pfizer to build its artificial intelligence-based platform. 

AbSci

Founded in 2011 in Washington, the bioengineering startup is building its protein printing platform technology, which consists of two parts: SoluPro, a synthetic biology platform that generates potential protein-based drugs from genetically engineered bacteria, and SoluPure, a purification platform that quickly extracts and purifies target cells. 

The platform relies on AI to help produce and test new drug candidates. In January 2021, the company acquired the Devonium Engine machine learning platform to continue work on its data processing technology. AbSci enables shorter preclinical timelines and even developed a COVID-19 antibody cell line in 10 days. 

Exscientia

The pharmaceutical startup, founded in 2012, has attracted a total of about $150 million in investment from Bristol-Myers Squibb, Evotec, and the Bill and Melinda Gates Foundation. The company developed the Centaur Chemist AI-based design platform, which uses machine learning methods to search for potential drugs. 

The AI platform studies the limited range of possible approaches used in traditional drug discovery by human chemists and improves with each successive iteration. Joint efforts of AI and humans have cut the average time to bring drugs to phase one from five years to 12 months. 

The company recently announced that its first AI-designed molecule for treating cancer using the body's own immune system will enter human clinical trials. 

PostEra

Founded in 2019, the company works at the intersection of medicinal chemistry and machine learning. The startup is currently a partner of Pfizer in developing machine learning for drug discovery. The PostEra platform is based on design strategies and chemical synthesis rules used by medicinal chemists and organic chemists to synthesize drug candidates. Pfizer plans to provide the startup with huge datasets to power the platform. 

Brightseed

Founded in 2017, the biotechnology startup works on discovering new health-beneficial molecules from the natural world. The company developed the Forager AI research platform to study the world of edible plants and foods in search of natural compounds that affect human health. Known as phytonutrients, many of these compounds are used as functional ingredients in foods, but their health effects are still not fully understood. 

To date, Brightseed has analyzed more than 700,000 bioactive compounds and discovered one natural food ingredient that may support liver health by addressing fatty liver disease, a condition associated with obesity. Forager will be used to expand the known set of food supplements.