NEWS COMMENTARY

Google DeepMind's hybrid deep learning approach again exceeds expectations in protein folding prediction competition

Published:
December 01, 2020
Services:
Consumer More...
Activities:
Research
Very important

While DeepMind's previous entry into the global CASP contest substantially improved upon other techniques to predict 3D protein structure from an amino acid sequence, the software still lacked practical applications, as most of its predictions weren't close enough to gold-standard experimental methods like X-ray crystallography to be useful. The company's entry this year crossed an important threshold by being on par with X-ray crystallography across more than two-thirds of the proteins in the competition. Clients should expect this to have a significant impact by reducing the time and cost to solve many protein structures; however, it will still be several more years before such approaches can consistently replace lab testing.

For the original news article, click here .