Atomic-level protein structure refinement using fragment guided molecular dynamics conformation sampling

By Jian Zhang, Yu Liang, Yang Zhang. 2011

329 citations recorded by the source corpus.

Summary One of critical difficulties of molecular dynamics (MD) simulations in protein structure refinement is that the physics-based energy landscape lacks of middle-range funnel to guide non-native conformations towards near-native states. We propose to use the target model as probe to identify fragmental analogs from PDB. The distance maps are then used to re-shape the MD energy funnel. The protocol was tested on 181 benchmarking and 26 CASP targets. It was found that structure models of correct folds with TM-score>0.5 can be often pulled closer to native with higher GDT-HA score; but improvement for the models of incorrect folds (TM-score<0.5) are much less pronounced. These data indicate that template-based fragmental distance maps essentially re-shaped the MD energy landscape from golf-course-like to funnel-like ones in the successfully-refined targets with a radius of TM-score~0.5. These results demonstrate a new avenue to improve high-resolution structures by combining knowledge-based template information with physics-based MD simulations.

Subfields: Materials Chemistry, Molecular Biology, Spectroscopy

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Materials Chemistry

1. Robin Pearce
2. Yang Zhang
3. Michael Feig
4. Chengxin Zhang
5. Jianyi Yang
6. S. Mortuza
7. Yang Li
8. Wei Zheng

In Molecular Biology

1. Yang Zhang
2. Robin Pearce
3. Chengxin Zhang
4. Yang Li
5. Wei Zheng
6. Eric W. Bell
7. Michael Feig
8. Jianyi Yang

In Spectroscopy

1. Robin Pearce
2. Yang Zhang
3. Jianyi Yang
4. Michael Feig
5. Chengxin Zhang
6. Wei Zheng
7. Yang Li
8. S. Mortuza

Citations in the corpus

• I-TASSER server: new development for protein structure and function predictions (2015, methodology): “In FG-MD (23), we relax the I-TASSER models by simulated annealing molecular dynamics simulations, where the distance maps from structure fragments detected by TM-align from the PDB library are used as spatial restraints.”
• I-TASSER-MTD: a deep-learning-based platform for multi-domain protein structure and function prediction (2022, methodology): “Finally, the full-length model with the lowest energy is selected for side-chain repacking by FASPR and FG-MD.”
• STRUM: structure-based prediction of protein stability changes upon single-point mutation (2016, methodology): “The structural model for the mutant sequence is reconstructed by SCWRL4 (Dunbrack and Cohen, 1997) by replacing the side-chain rotamers according to the mutated amino acids, followed by atomic-level FG-MD refinement simulations (Zhang et al. , 2011).”
• Template-based protein structure prediction in CASP11 and retrospect of I-TASSER in the last decade (2015, methodology): “As outlined in Method, both I-TASSER and QUARK simulations are built on the reduced models; the atomic details are quickly added by REMO30 after the simulations, which are then refined by FG-MD.”
• Folding non-homologous proteins by coupling deep-learning contact maps with I-TASSER assembly simulations (2021, methodology): “…(Li and Zhang, 2009) https://zhanglab.ccmb.med.umich.edu/ REMO/ FASPR (Huang et al., 2020) https://zhanglab.ccmb.med.umich.edu/ FASPR/ FG-MD (Zhang et al., 2011) https://zhanglab.ccmb.med.umich.edu/ FG-MD/ ThreaDom (Xue et al., 2013) https://zhanglab.ccmb.med.umich.edu/ ThreaDom/ DEMO (Zhou…”
• Toward the solution of the protein structure prediction problem (2021, methodology): “Nevertheless, molecular dynamics has remained a popular tool to study protein motion (128, 131) and for full-atom re fi nement of protein structures (132 – 134).”
• Detecting distant-homology protein structures by aligning deep neural-network based contact maps (2019, methodology): “Following the simulations, the decoys from the simulation trajectories are clustered by SPICKER [51], where the conformations that are closest to cluster centroids are selected and refined by FG-MD [52] to generate the final top 5 structure models.”
• Using iterative fragment assembly and progressive sequence truncation to facilitate phasing and crystal structure determination of distantly related proteins. (2016, methodology): “The models obtained in the second-round refinement simulations are further refined using fragment-guided molecular-dynamics all-atom simulations (Zhang et al., 2011).”
• Computational protein structure refinement: Almost there, yet still so far to go (2017, methodology): “In particular, the hydrogen bond network that is key for secondary structure formation may be examined and reoptimized 55,93,106 or structural elements of a given model such as secondary structure elements or loops may be resampled extensively to fi nd better conformations.”