2018

December

1 events

Editors' summary

AlphaFold takes first place in protein prediction

On 2 December, DeepMind's AlphaFold took first place at CASP13 in protein structure prediction, producing the most accurate predictions on 25 of 43 targets.

A team known for its results in games reaching into a scientific problem drew attention. The accuracy was not yet a substitute for experiment, and the next version was awaited.

This block is written by the editors. It is kept separate from the sourced record below.

Record1 events
  1. 2
    ResearchGoogle / John Jumper / Demis Hassabis

    DeepMind's AlphaFold wins the CASP13 protein folding contest

    DeepMind announced that AlphaFold had placed first at CASP13, the international competition in protein structure prediction, held in Cancún, Mexico. It beat 98 competing systems and predicted the structure of 25 of 43 proteins with high accuracy, the product of two years of work. Predicting three-dimensional structure from an amino acid sequence had been a 50-year-old problem, and the result showed AI could outperform human-designed methods on a central question in basic science. It sits outside the generative AI lineage, but it was one of Google's answers to what AI is for. Its successor, AlphaFold2, would compete at CASP14 two years later.