Artificial Intelligence for Drug Discovery
We design new molecules for the hardest targets. Our discoveries help our partners deliver better medicines faster.
Enabling Success
Our machine learning has improved hit rates by up to 10,000x and we’re able to deliver accuracy comparable to wet lab experiments. We screen for potency, selectivity, and polypharmacology, and guard against off-target toxicity.
Unparalleled Speed
Atomwise delivers results 100 times faster than ultra high throughput screening. Our deep convolutional neural network, AtomNet, screens between 10 and 20 million compounds a day.
Tackling Hard Problems
We can work on the hardest targets including previously undrugged proteins, and we've delivered in vivo success in blocking protein-protein interactions and targets in the central nervous system.
Working to Fight an Epidemic
Ebola virus’ high lethality has left thousands orphaned since 1976. From a field of thousands of approved medicines, Atomwise discovered a drug candidate with no previous antiviral application that blocked Ebola infectivity across virus strains from multiple epidemics.
Our team of medicinal chemists, structural biologists,
and machine learning experts are world class.
Atomwise Raises $45 Million Series A
Atomwise’s Series A marks the culmination of two years of unprecedented business growth. Atomwise has struck partnerships with four of the top-ten U.S. pharma companies, multiple biotech firms, and over forty major research universities.
Left: An untreated EAE mouse. Right: An EAE mouse treated with an Atomwise-predicted small molecule therapy.
Computing new treatment for multiple sclerosis
We were able to inhibit a protein-protein interaction in the central nervous system for a multiple sclerosis project, with a molecule that is orally active in mouse models at 3 nM/g.
Powered by Our AtomNet Technology
AtomNet is the first deep learning neural network for structure based drug design and discovery. Its speed and accuracy make it the most advanced technology for small molecule binding affinity prediction.
Visualization of AtomNet searching for small molecules to bind to the human ROCK2 kinase.