In 2018, Google’s DeepMind made waves when its new neural network AI program, AlphaFold 1, placed first at the Critical Assessment of Structure Prediction (CASP) competition. CASP, a global experiment to refine the methods and models for protein structure prediction, is known as the “world championship” in the field of protein research and takes place every two years. Protein structure prediction requires scientists to infer the three-dimensional structure of a protein based on its amino acid sequence. Knowing a protein’s structure is extremely valuable in medicine because of its importance to drug design.
In 2020, DeepMind’s updated AlphaFold 2 won CASP again, cementing its place as an entirely novel and groundbreaking feat for protein structure prediction. By 2022, DeepMind did not even enter CASP, but most of the competitors relied on or incorporated AlphaFold’s tools.
Proteins: The Building Blocks of Life
Proteins are highly complex, made up of strings of amino acids that twist, tangle, and fold. Because a protein’s structure is inseparable from its function, knowing its structure can help explain the protein itself. Before AlphaFold, scientists could not tackle this “protein folding problem” without years of effort and work. There are around 20,000 proteins in the human body and 200 million proteins outside of the human body. These proteins are the building blocks of life. They provide structural support, binding cells together into tissues, muscles, and tendons. They enable immune responses by fighting foreign viruses and bacteria through protein antibodies. They bind to DNA and regulate gene expression. They serve as intravenous transportation vehicles by binding to specific molecules, such as cholesterol and triglycerides. And these functions are nowhere near an exhaustive list of what proteins do in living organisms.
A Gamechanger for Longevity?
“AlphaFold can predict the shape of proteins to within the width of an atom. The breakthrough will help scientists design drugs and understand disease,” wrote an author at the MIT Technology Review. Incredibly, DeepMind publicly released AlphaFold’s code and all the protein structures it uncovered in an open database. As of 2022, this treasure trove of scientific data covers over 200 million protein structures and details the structure of almost every protein with a genetic sequence known to science, including all 20,000 proteins found in the human body (also known as the human proteome). Importantly, AlphaFold 2 includes a confidence measure for every prediction, so that scientists have a clear idea of how certain they should be about a structure’s accuracy.
Hailed in Forbes as “the most important achievement in AI – ever,”AlphaFold’s implications for longevity science are enormous. AlphaFold is being used to accelerate drug discovery and basic research, from developing a malaria vaccine to fighting antibiotic resistance to exploring the role of certain proteins in Parkison’s disease protection. Lifespan and healthspan can both be improved by therapeutics that use AlphaFold. To put it in perspective,
“Using AlphaFold, a team at the University of Colorado Boulder was able to pinpoint a particularly tricky bacterial protein structure, a discovery that will aid their efforts to combat antibiotic resistance, a looming public health crisis. The Boulder team had spent years unsuccessfully trying to determine this protein’s structure; with AlphaFold, they learned it in 15 minutes.
Nonetheless, AlphaFold itself did not discover any new lifesaving drugs. Most of its predicted structures are not computed with absolute certainty, and biologists developing new pharmaceuticals still have to engage in difficult, complicated lab work to put AlphaFold’s insights to good use. At the end of the day, AlphaFold did not solve drug discovery or longevity science, but it certainly accelerated them.