Using a conformation-dependent stereochemical library improves crystallographic refinement of proteins

The major macromolecular crystallographic refinement packages restrain models to ideal geometry targets defined as single values that are independent of molecular conformation. However, ultrahigh-resolution X-ray models of proteins are not consistent with this concept of ideality and have been used to develop a library of ideal main-chain bond lengths and angles that are parameterized by the φ/ψ angle of the residue [Berkholzet al.(2009),Structure,17, 1316–1325]. Here, it is first shown that the new conformation-dependent library does not suffer from poor agreement with ultrahigh-resolution structures, whereas current libraries have this problem. Using theTNTrefinement package, it is then shown that protein structure refinement using this conformation-dependent library results in models that have much better agreement with library values of bond angles with little change in theRvalues. These tests support the value of revising refinement software to account for this new paradigm.

Download Full-text

Accurate bond and angle parameters for X-ray protein structure refinement

Acta Crystallographica Section A Foundations of Crystallography ◽

10.1107/s0108767391001071 ◽

1991 ◽

Vol 47 (4) ◽

pp. 392-400 ◽

Cited By ~ 1756

Author(s):

R. A. Engh ◽

R. Huber

Keyword(s):

Protein Structure ◽

Structure Refinement ◽

X Ray ◽

Protein Structure Refinement

Download Full-text

Improved protein structure refinement guided by deep learning based accuracy estimation

Nature Communications ◽

10.1038/s41467-021-21511-x ◽

2021 ◽

Vol 12 (1) ◽

Author(s):

Naozumi Hiranuma ◽

Hahnbeom Park ◽

Minkyung Baek ◽

Ivan Anishchenko ◽

Justas Dauparas ◽

...

Keyword(s):

Deep Learning ◽

Protein Structure ◽

Structure Refinement ◽

Global Energy ◽

Accuracy Estimation ◽

X Ray ◽

Learning Framework ◽

Protein Structure Refinement ◽

Protein Models ◽

Multiple Stages

AbstractWe develop a deep learning framework (DeepAccNet) that estimates per-residue accuracy and residue-residue distance signed error in protein models and uses these predictions to guide Rosetta protein structure refinement. The network uses 3D convolutions to evaluate local atomic environments followed by 2D convolutions to provide their global contexts and outperforms other methods that similarly predict the accuracy of protein structure models. Overall accuracy predictions for X-ray and cryoEM structures in the PDB correlate with their resolution, and the network should be broadly useful for assessing the accuracy of both predicted structure models and experimentally determined structures and identifying specific regions likely to be in error. Incorporation of the accuracy predictions at multiple stages in the Rosetta refinement protocol considerably increased the accuracy of the resulting protein structure models, illustrating how deep learning can improve search for global energy minima of biomolecules.

Download Full-text