DeepMind AI hits new milestone in protein folding prediction

A new AI system from DeepMind excels at protein structure modeling.

Artificial intelligence (AI) has the potential to revolutionize the world in a way that few technologies can. It has applications in almost every industry and can make life better for everyone.

Some industries have even more to gain than others. For instance, scientists can use AI to perform tasks that once took years to finish in a matter of minutes.

That’s exactly what researchers are doing with AI tech from DeepMind, a well-known subsidiary of Alphabet. The company recently announced that its AlphaFold system has hit a major milestone. It solved a protein folding challenge that has eluded scientists for decades.

Manage your supply chain from home with Sourcengine

This signals that AI-based protein structure prediction is almost ready to make a major impact in a variety of scientific fields.

Molecular Understanding

From drug development to disease research, understanding how proteins fold into their unique shapes is a key part of the process. Unfortunately, trying to determine the shape of proteins with traditional means takes a long, long time. It also requires tremendous amounts of computing power—something that most labs don’t have access to.

DeepMind’s AlphaFold AI has shown that it is capable of predicting protein structures in as little as a few days. That is a major achievement for the system and carries massive implications for the scientific community.

In July, The Burn-In highlighted IBM’s World Community Grid project. It, like other crowdsourced computing efforts, seeks to utilize resources from idle machines around the world to carry out protein structure modeling. Until now, that has been one of the best ways to do so without the use of a supercomputer.

DeepMind’s AI-based approach is a bit different. It uses an “attention-based neural network system” to refine its predictions in real-time. The AI is able to focus on specific inputs to boost its efficiency, TechCrunch notes.

The results speak for themselves.

AlphaFold is able to correctly identify the shape of proteins based on a strand of amino acids. The system is accurate to within the width of an atom. Although that isn’t yet good enough to eliminate reliance on human researchers, it does greatly reduce the time it takes to identify protein shapes.

Faster Discovery

As mentioned, this breakthrough is a big deal for the scientific community. Reducing the time it takes to accurately predict protein structures helps scientists unravel complex disease processes and identify promising new drugs.

With a better understanding of proteins, scientists will be able to create new treatments for things like cancer, AIDS, and even future pandemics. Likewise, a better understanding of these conditions makes it possible to devise prevention strategies and early interventions.

The COVID-19 pandemic has made it clear that the world is capable of responding to serious threats. With the help of AI, humanity’s next response can be much faster.

The promising AlphaFold data still needs to be peer-reviewed and published. However, it appears that DeepMind’s AI system is both highly accurate and efficient. The company says that it is “optimistic about the impact AlphaFold can have on biological research and the wider world.”

This AI is worth watching in the coming years as it begins to play a larger role in scientific discovery.


Please enter your comment!
Please enter your name here