Atomwise’s deep learning technology could speed up drug discovery

Image: Drew Hays - Unsplash
Ingrid Fadelli

The company just raised $45m in Series A funding to develop its AI technology for pharmaceutical research

Atomwise, a San Francisco-based company that develops Artificial Intelligence (AI) technology for drug discovery, has recently raised $45 million in Series A funding, in a round led by Mosanto Growth Ventures, Data Collective (DCVC), and B Capital Group.

The AI company is developing AI technology that could drastically reduce time, money, and resources invested in trying to identify new, effective drug treatments, as well as safer and more effective agricultural pesticides.

Atomwise’s new investors include Baidu Ventures, Tencent, and Dolby Family Ventures, but also participating in the round are returning investors such as Y Combinator, Khosla Ventures, and DFJ.

Said Monsanto Growth Ventures partner Kiersten Stead in a press statement:

“We chose to invest based on the impressive results we saw from Atomwise in our own hands. Atomwise was able to find promising compounds against crop protection targets that are important areas of focus for agrochemical R&D.”

Since the introduction of machine learning strategies, the field of AI has made huge leaps forward, with researchers worldwide now developing neural networks that can analyze huge amounts of data in very limited time; making highly accurate predictions and speeding up research within several fields.

Atomwise’s technology could play a part in the discovery of new effective chemical compounds for both medical and agricultural sectors.

AI that could speed up drug discovery

Founded in 2012, Atomwise has already raised a total of $51 million in funding and is currently working on more than 50 molecular discovery programs.

The technology developed by the company, AtomNet, is a deep learning algorithm for the discovery of new small molecules. It works by analyzing simulations of molecules and identifying the most effective, drastically reducing the time human researchers spend on synthetizing and testing different compounds.

Atomwise’s client portfolio includes four of the biggest pharmaceutical companies in the US, including Merck and Monsanto, as well as 40 renowned universities, such as Harvard, Duke, and Stanford.

The company says that AtomNet is currently screening more than 10 million compounds a day, trying to discover new effective drugs that could treat specific medical conditions.

In an e-mail to Tech Crunch, Atomwise’s CEO Abraham Haifets said that the company’s mission is “to become one of the most prolific and diverse life science research groups in the world, working at a scale that is truly unprecedented.”

He commented on the recently attained investment saying:

“This is a large Series A and we will use these resources to grow our technical and business organization. We may eventually find ourselves simulating hundreds of millions of compounds per day. The ultimate upshot is more shots on goal for the many diseases that urgently need new treatments.”

Researchers at Atomwise trained AtomNet to analyze molecules and predict their actual effects on the human body, including potential side effects, toxicity issues, and their overall efficacy.

There are two main steps in drug discovery; the biological and the chemical one. Scientists working on biology try to decide what disease protein is the best target for a new drug, while those working on chemistry try to identify a non-toxic molecule that could hit this desired target.

“Atomwise is focused on these chemistry problems,” explains Haifets. “Specifically, Atomwise invented the use of deep neural networks for structure-based drug design.”

New AI-fueled research practices

New AI technologies such as AtomNet are gradually re-shaping the way in which scientists carry out research, enabling faster discoveries and more productive research practices.

Atomwise’s deep learning algorithm could dramatically speed up pharmaceutical research, which is currently among the most time-consuming and costly scientific practices.

In the e-mail to TechCrunch, Heifets highlighted that lead optimization has always been the most expensive step in the pharmaceutical pipeline and that it often has a very high failure rate, with around two-thirds of projects failing to reach clinics.

Technologies such as AtomWise could drastically accelerate this process, allowing researchers to identify effective compounds faster without having to carry out enormous amounts of tests.

“With our partners, Atomwise has brought the power of artificial intelligence to breakthrough research on deadly viruses, several forms of cancer, neurodegenerative diseases, metabolic diseases, life-threatening bacteria, endemic parasites, and crop-blighting fungi in agriculture,” said Dr. Izhar Wallach, Co-Founder and CTO of Atomwise, in a Press Release posted by Business Wire. “With this funding, Atomwise is ready to help hundreds of organizations discover compounds that could become tomorrow’s blockbusters.”

Atomwise was one of the first companies to start developing AI for pharmaceutical research, but several others have now followed, including Reverie Labs, Recursion Pharmaceuticals, Cyclica, and Benevolent AI, a startup that Clique reported on in late 2017.

We are clearly approaching an AI-fueled transformation that could, among other things, result in faster and more effective scientific research practices, potentially leading to an exciting wave of new ground-breaking discoveries.

Start typing and press Enter to search