May 25, 2022
New Delhi
science

AI Deep Mind Solved the 50-Year Old Problem in Biology: The Making of a Scientific Breakthrough

AI Deep Mind Solved the 50-Year Old Problem: The Making of a Scientific Breakthrough

Animated image shows spinning protein structures

Life on Earth relies on minute machines called proteins that are essential to everything from stabilizing the structure of every cell to reading the genetic code and transporting oxygen through the bloodstream. Robert F of Science. Through careful laboratory work, scientists have come up with the exact 3D shape of about 170,000 proteins, according to the Service report, but there are at least 200 million other proteins.

According to Will Douglas Heaven’s MIT Tech Review report, researchers have been trying to find effective ways to estimate protein shape since at least the 1970s. According to the New York Times Kid Metz report, AI company DeepMind, now owned by the same company that owns Google, has developed a tool that can predict the 3D shape of most proteins with similar results to lab experiments. Did. Times. According to Nature’s Ewen Callaway, lab experiments can take years to study protein structure, but DeepMind, called AlphaFold, can reach structure in days, developing medical and bioengineering. Helps speed up your research.

Molecular biologists want to know the structure of proteins because the shape of the molecule determines what it can do. For example, if a protein is harmful to the body, scientists can study its structure and find another protein that is suitable for it. Such as a piece of a puzzle to neutralize it. AlphaFold can speed up this process.

“This will allow a new generation of molecular biologists to ask more advanced questions,” Andre Rubas, an evolutionary biologist at the Max Planck Institute, told Nature. “We need to think a little more and sample a little more.”

DeepMind tested AlphaFold in a biennial challenge called the Important Evaluation of Structure Prediction (CASP). At CASP, Lupas was the judge. CASP provides a framework for developers to test protein prediction software. It has been functioning since 1994. But recent increases in machine learning in protein structure prediction have taken participants to a new level. According to a scientific journal, AlphaFold participated for the first time last year, scoring about 15% higher than other entries. This year, a new algorithm allowed AlphaFold to win the competition.

Industry Reaction to this Success

Leave feedback about this

  • Quality
  • Price
  • Service

PROS

+
Add Field

CONS

+
Add Field
Choose Image
Choose Video
X