AI Creates Stunning Advance In Fight Against Disease

Back in the early 2000s, a popular way for the average person to help in the fight against disease was to run a program called Folding At Home which used idle CPU time to discover the shape of proteins, helping uncover ways to beat diseases such as Cancer, Parkinson’s and COVID. With over 200,000,000 different proteins to examine, this distributed computing approach was used due to the sheer complexity of the problem at hand. Now, AI has come along and uncovered a new way to solve this problem, decades earlier than expected. 

Google’s DeepMind has created a new AI model called AlphaFold which can solve these complex problems in days, instead of months or years, speeding the time it takes to understand how these proteins affect diseases of the human body. This program is the result of a collection of scientists and engineers that have been studying the folding of proteins since 1994, as part of the 14th Community Wide Experiment on the Critical Assessment of Techniques for Protein Structure Prediction (CASP14). 

“Proteins are extremely complicated molecules, and their precise three-dimensional structure is key to the many roles they perform … Even tiny rearrangements of these vital molecules can have catastrophic effects on our health, so one of the most efficient ways to understand disease and find new treatments is to study the proteins involved.”
-Dr John Moult, chair of CASP14.

DeepMind’s AlphaFold has been shown to be able to predict the shape of two-thirds of proteins it processed with high accuracy. The DeepMind team has more planned for the new machine learning model, including finding ways to better understand how proteins interact with other proteins and even DNA.