Dataset Information

Protein tertiary structure prediction and refinement using deep learning and Rosetta in CASP14.

ABSTRACT: The trRosetta structure prediction method employs deep learning to generate predicted residue-residue distance and orientation distributions from which 3D models are built. We sought to improve the method by incorporating as inputs (in addition to sequence information) both language model embeddings and template information weighted by sequence similarity to the target. We also developed a refinement pipeline that recombines models generated by template-free and template utilizing versions of trRosetta guided by the DeepAccNet accuracy predictor. Both benchmark tests and CASP results show that the new pipeline is a considerable improvement over the original trRosetta, and it is faster and requires less computing resources, completing the entire modeling process in a median < 3 h in CASP14. Our human group improved results with this pipeline primarily by identifying additional homologous sequences for input into the network. We also used the DeepAccNet accuracy predictor to guide Rosetta high-resolution refinement for submissions in the regular and refinement categories; although performance was quite good on a CASP relative scale, the overall improvements were rather modest in part due to missing inter-domain or inter-chain contacts.

SUBMITTER: Anishchenko I

PROVIDER: S-EPMC8616808 | biostudies-literature | 2021 Dec

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Protein tertiary structure prediction and refinement using deep learning and Rosetta in CASP14.

Anishchenko Ivan I Baek Minkyung M Park Hahnbeom H Hiranuma Naozumi N Kim David E DE Dauparas Justas J Mansoor Sanaa S Humphreys Ian R IR Baker David D

Proteins 20210817 12

The trRosetta structure prediction method employs deep learning to generate predicted residue-residue distance and orientation distributions from which 3D models are built. We sought to improve the method by incorporating as inputs (in addition to sequence information) both language model embeddings and template information weighted by sequence similarity to the target. We also developed a refinement pipeline that recombines models generated by template-free and template utilizing versions of tr ...[more]

PMID: 34331359

Similar Datasets

Project description:Accurate prediction of residue-residue distances is important for protein structure prediction. We developed several protein distance predictors based on a deep learning distance prediction method and blindly tested them in the 14th Critical Assessment of Protein Structure Prediction (CASP14). The prediction method uses deep residual neural networks with the channel-wise attention mechanism to classify the distance between every two residues into multiple distance intervals. The input features for the deep learning method include co-evolutionary features as well as other sequence-based features derived from multiple sequence alignments (MSAs). Three alignment methods are used with multiple protein sequence/profile databases to generate MSAs for input feature generation. Based on different configurations and training strategies of the deep learning method, five MULTICOM distance predictors were created to participate in the CASP14 experiment. Benchmarked on 37 hard CASP14 domains, the best performing MULTICOM predictor is ranked 5th out of 30 automated CASP14 distance prediction servers in terms of precision of top L/5 long-range contact predictions (i.e. classifying distances between two residues into two categories: in contact (< 8 Angstrom) and not in contact otherwise) and performs better than the best CASP13 distance prediction method. The best performing MULTICOM predictor is also ranked 6th among automated server predictors in classifying inter-residue distances into 10 distance intervals defined by CASP14 according to the precision of distance classification. The results show that the quality and depth of MSAs depend on alignment methods and sequence databases and have a significant impact on the accuracy of distance prediction. Using larger training datasets and multiple complementary features improves prediction accuracy. However, the number of effective sequences in MSAs is only a weak indicator of the quality of MSAs and the accuracy of predicted distance maps. In contrast, there is a strong correlation between the accuracy of contact/distance predictions and the average probability of the predicted contacts, which can therefore be more effectively used to estimate the confidence of distance predictions and select predicted distance maps. The software package, source code, and data of DeepDist2 are freely available at https://github.com/multicom-toolbox/deepdist and https://zenodo.org/record/4712084#.YIIM13VKhQM.

Project description:Cryo-EM has become one of the prime methods for protein structure elucidation, frequently yielding density maps with near-atomic or medium resolution. If protein structures cannot be deduced unambiguously from the density maps, computational structure refinement tools are needed to generate protein structural models. We have previously developed an iterative Rosetta-MDFF protocol that used cryo-EM densities to refine protein structures. Here we show that, in addition to cryo-EM densities, incorporation of other experimental restraints into the Rosetta-MDFF protocol further improved refined structures. We used NMR chemical shift (CS) data integrated with cryo-EM densities in our hybrid protocol in both the Rosetta step and the molecular dynamics (MD) simulations step. In 15 out of 18 cases for all MD rounds, the refinement results obtained when density maps and NMR chemical shift data were used in combination outperformed those of density map-only refinement. Notably, the improvement in refinement was highest when medium and low-resolution density maps were used. With our hybrid method, the RMSDs of final models obtained were always better than the RMSDs obtained by our previous protocol with just density refinement for both medium (6.9 Å) and low (9 Å) resolution maps. For all the six test proteins with medium resolution density maps (6.9 Å), the final refined structure RMSDs were lower for the hybrid method than for the cryo-EM only refinement. The final refined RMSDs were less than 1.5 Å when our hybrid protocol was used with 4 Å density maps. For four out of the six proteins the final RMSDs were even less than 1 Å. This study demonstrates that by using a combination of cryo-EM and NMR restraints, it is possible to refine structures to atomic resolution, outperforming single restraint refinement. This hybrid protocol will be a valuable tool when only low-resolution cryo-EM density data and NMR chemical shift data are available to refine structures.

Dataset Information

Protein tertiary structure prediction and refinement using deep learning and Rosetta in CASP14.

Publications

Protein tertiary structure prediction and refinement using deep learning and Rosetta in CASP14.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets