Charm Therapeutics Applies AI to Complex Protein Interactions, Blocking $50M Round

- Advertisement -


The world of AI-assisted drug discovery continues to expand as the power of machine learning grows. One approach that seemed unthinkable just a few years ago is to model the complex interactions of two interconnected molecules, but this is exactly what drug developers need to be aware of, and this is exactly what Charm Therapy aims to do with its DragonFold platform.

- Advertisement -

Proteins do almost everything that needs to be done in your body and are the most common target for drugs. And in order to create an effect, you must first understand this purpose, in particular, how the chain of amino acids that make up a protein “folds” under different circumstances.

- Advertisement -

In the recent past, this was often done with complex and time-consuming X-ray crystallography, but recently it has been shown that machine learning models such as AlphaFold as well as RoseTTAFold capable of producing the same good results, but in seconds rather than weeks or months.

The next problem is that even if we know how a protein folds under the most common conditions, we do not know how it can interact with other proteins, let alone new molecules designed specifically to bind to them. When a protein encounters a compatible binder or ligand, it can completely transform, as small changes can cascade and reconfigure its entire structure – in life, this results in the protein opening a passageway into the cell or exposing a new surface that activates other proteins, and so on. Further.

- Advertisement -

“This is really where we innovated: we created DragonFold, which is the first protein-ligand co-folding algorithm,” said Lusk Aitani, CEO and co-founder of Charm Therapeutics.

“Designing drugs that bind very tightly and selectively to the disease-causing protein of interest (that is, avoid binding to other similar proteins that are essential for normal human function) is paramount,” he explained. “This is easiest to do when you know exactly how these drugs bind to the protein (the exact three-dimensional shape of the ligand bound to the disease-causing protein). This allows for precise modifications to the ligand so that it can bind more tightly and selectively.”

You can see a picture of this situation at the beginning of the article: a small green molecule and a purple protein fit together in a very specific way that is not always intuitive or easily predictable. Efficient and efficient modeling of this process helps to scan billions of molecules, similar to earlier processes that identified drug candidates, but go further and reduce the need to experimentally test whether they interact properly.

To do this, Aitani turned to David Baker, the developer of the RoseTTAFold algorithm among many others and the head of influential laboratory at the University of Washingtonto become its co-founder. Baker is well known in academia and industry as one of the leading researchers in this field, and he has published numerous papers on the subject.

Charm Therapeutics co-founders Lash Aitani (left) and David Baker.

Shortly after algorithms were shown to be able to predict the structure of proteins based on their sequence, Baker established that they can also “hallucinate” new proteins which acted as expected in vitro. Soros, he is clearly at the forefront. And he won the $3 million Breakthrough Prize in 2020. — definitely up to becoming a technical co-founder. Aitani also proudly noted the presence of DeepMind veteran Sergey Bartunov as Director of Artificial Intelligence and former Head of Pharmaceutical Research Sarah Skerratt as Head of Drug Development.

The $50M Round A was led by F-Prime Capital and OrbiMed, with contributions from General Catalyst, Khosla Ventures, Braavos and Axial. While such large numbers are not uncommon for software startups, it should be noted that Charm doesn’t stop at creating capabilities to describe these protein-ligand interactions.

Early stage funding from the company was used to create the model, but they are now moving on to the next step: positive identification of effective drugs.

“We have the initial version [of the model] ready and it has been tested in silico,” Aitani said. “In the coming quarters, we will test this experimentally. Please note that the “product” will primarily be for internal use to help our own scientists discover potential drugs for which we own 100% rights.”

Typically, the testing process involves laboratory screening of thousands upon thousands of candidate molecules, but if it works as advertised, DragonFold should cut that number down considerably. This means that a relatively small lab with a relatively small budget can create a drug that a few years ago could have cost hundreds of millions of dollars from a large pharmaceutical company.

Given the new drug’s profit profile, it’s not surprising that the company has attracted this kind of investment: several tens of millions is a drop in the bucket compared to the R&D budget of any major biotech research company. All it takes is one hit and they laugh. It still takes a while, but AI drug discovery also shortens the timeline – so expect to hear about their first candidates sooner rather than later.


Credit: techcrunch.com /

- Advertisement -

Stay on top - Get the daily news in your inbox

DMCA / Correction Notice

Recent Articles

Related Stories

Stay on top - Get the daily news in your inbox