MIT researchers developed the EquiBind deep learning model, which is 1200 times faster than its peers, binds molecules to proteins to create drugs.
Before starting drug development, researchers must first find molecules that can “dock” with specific target proteins. However, this process requires significant financial and computational resources. Moreover, it takes decades to develop and test a new drug, and 90% of discoveries fail, scientists said.
According to study lead author Hannes Stark, existing binding methods ligand with protein are reminiscent of “trying to insert a key into a lock with a lot of keyholes.”
“Typical models take a lot of time and evaluate each “fit” before choosing the best option. In contrast, EquiBind directly predicts the exact location of a key in one step without prior knowledge of the target pocket of the protein, which is known as “blind matching,” the scientist said.
According to the researchers, the model has built-in geometric reasoning that helps it learn the underlying physics of molecules and make better predictions when faced with new, unknown data.
Relay Therapeutics data director Pat Walters suggested that scientists test the system on an existing drug and protein used in the treatment of lung cancer, leukemia and tumors of the gastrointestinal tract. According to him, the algorithm successfully bound ligands to proteins, which traditional methods failed to do.
“EquiBind offers a unique solution to the docking problem that includes both pose prediction and anchor location identification,” Walters says.
The researchers plan to present the algorithm at the International Conference on Machine Learning (ICML). Stark stated that the team intends to collect more feedback on the system from experts in the industry in order to improve it.
Recall that in March, an artificial intelligence algorithm deduced more than 40,000 variants of chemical weapons in six hours.
In November 2021, Alphabet Holding founded a new AI-assisted drug discovery company.
In July, artificial intelligence from DeepMind modeled 20,000 human protein structures.
Subscribe to Cryplogger news in Telegram: Cryplogger AI – all the news from the world of AI!
Found a mistake in the text? Select it and press CTRL+ENTER
MIT researchers developed the EquiBind deep learning model, which is 1200 times faster than its peers, binds molecules to proteins to create drugs.
Before starting drug development, researchers must first find molecules that can “dock” with specific target proteins. However, this process requires significant financial and computational resources. Moreover, it takes decades to develop and test a new drug, and 90% of discoveries fail, scientists said.
According to study lead author Hannes Stark, existing binding methods ligand with protein are reminiscent of “trying to insert a key into a lock with a lot of keyholes.”
“Typical models take a lot of time and evaluate each “fit” before choosing the best option. In contrast, EquiBind directly predicts the exact location of a key in one step without prior knowledge of the target pocket of the protein, which is known as “blind matching,” the scientist said.
According to the researchers, the model has built-in geometric reasoning that helps it learn the underlying physics of molecules and make better predictions when faced with new, unknown data.
Relay Therapeutics data director Pat Walters suggested that scientists test the system on an existing drug and protein used in the treatment of lung cancer, leukemia and tumors of the gastrointestinal tract. According to him, the algorithm successfully bound ligands to proteins, which traditional methods failed to do.
“EquiBind offers a unique solution to the docking problem that includes both pose prediction and anchor location identification,” Walters says.
The researchers plan to present the algorithm at the International Conference on Machine Learning (ICML). Stark stated that the team intends to collect more feedback on the system from experts in the industry in order to improve it.
Recall that in March, an artificial intelligence algorithm deduced more than 40,000 variants of chemical weapons in six hours.
In November 2021, Alphabet Holding founded a new AI-assisted drug discovery company.
In July, artificial intelligence from DeepMind modeled 20,000 human protein structures.
Subscribe to Cryplogger news in Telegram: Cryplogger AI – all the news from the world of AI!
Found a mistake in the text? Select it and press CTRL+ENTER
