To expedite drug development, Fujitsu uses generative AI

Fujitsu and Riken are developing AI to predict drug's target protein location 10x faster than current methods.

Fujitsu and Riken have used generative artificial intelligence to recreate proteins as three-dimensional structures in motion using images of the proteins captured with advanced cryo-electron microscopy.

As early as fiscal 2024, trials will get underway. This month, Fujitsu initiated contact with the approximately ninety members of the Life Intelligence Consortium, comprising universities and pharmaceutical companies.

Antibodies that bind to particular proteins that induce illness and other conditions are used in drug administration; these compounds aid in stopping viruses from invading cells or attacking cancer.

The process of predicting the form and motion of the protein that a particular medicine targets can be sped up thanks to this new technology. In a ribosomal protein experiment, a procedure that required experts for a whole day was shortened to two hours.

To forecast a protein's 3D structure, researchers have traditionally examined a huge number of electron microscope pictures, and it has been challenging to anticipate a protein's movements with precision. The work takes months because of the physical restrictions on what can be inspected at a time.

The long-term goal is to create artificial intelligence (AI) that can use genetic information to predict atomic-level protein movements.

In roughly five years, this could be used in the drug discovery industry, according to Takashi Kato of Fujitsu's Artificial Intelligence Laboratory.

Early on, thorough analysis with generative AI would allow resources to be concentrated on medication candidates that show promise, raising the likelihood that research will be successful. This might hasten the development of medications to treat infectious illnesses.