'It will change everything': DeepMind's AI makes gigantic leap in solving protein structures.

  title={'It will change everything': DeepMind's AI makes gigantic leap in solving protein structures.},
  author={Ewen Callaway},
  • E. Callaway
  • Published 30 November 2020
  • Medicine, Engineering
  • Nature
Science News for Theological Study: Machine Learning Unravels the Protein Folding Knot
Back in November 2020, in the depths of Covid-19 lockdowns, news broke that must count as one of the the science and technology stories of the decade, perhaps even of the century. A machine learning
The breakthrough in protein structure prediction
The path to CASP14 is reviewed, the method employed by AlphaFold2 to the extent revealed is outlined, and the implications of this breakthrough for the life sciences are discussed.
The trRosetta server for fast and accurate protein structure prediction.
The trRosetta server distinguishes itself from other similar structure prediction servers in terms of rapid and accurate de novo structure prediction, and homologous templates are used as additional inputs to the network automatically to take advantage of homology modeling.
Toward the solution of the protein‐structure prediction problem
The problem can be solved in principle by TBM if fold‐recognition algorithms could identify the best structural templates from the PDB, and results in the recent community‐wide blind CASP experiments are discussed, showing that new approaches combining ab initio folding and deep neural‐network contact and distance predictions, can result in consistent and successful folding of large proteins with complicated shapes and topologies.
The structural coverage of the human proteome before and after AlphaFold
Overall, the results show that the sequence-structure gap of human proteins has almost disappeared, an outstanding success of direct consequences for the knowledge on the human genome and the derived medical applications.
Evaluating the Reliability of AlphaFold 2 for Unknown Complex Structures with Deep Learning
AlphaFold-Eva is developed, a deep learning-based method that learns geometry information from complex structures to evaluate AlphaFold 2 and suggests that the reliability of predicted structures can be directly learned from the intrinsic structural information itself.
Deep learning techniques have significantly impacted protein structure prediction and protein design.
AlphaDesign: A graph protein design method and benchmark on AlphaFoldDB
A new method called ADesign is proposed to improve accuracy by introducing protein angles as new features, using a simplified graph transformer encoder (SGT), and proposing a confidence-aware protein decoder (CPD) based on AlphaDesign.
One Plus One Makes Three: Triangular Coupling of Correlated Amino Acid Mutations.
This work demonstrates that calculations based on the first-principles of statistical mechanics are capable of capturing the effects of nonadditivities in protein mutations and identifies thermodynamic couplings that cover the short-range as well as previously unknown long-range correlations.