Smooth orientation-dependent scoring function for coarse-grained protein quality assessment

Making Use of Averaging Methods in MODELLER for Protein Structure Prediction

Recent advances in protein structure prediction, driven by AlphaFold 2 and machine learning, demonstrate proficiency in static structures but encounter challenges in capturing essential dynamic features crucial for understanding biological function. In this context, homology-based modeling emerges as a cost-effective and computationally efficient alternative. The MODELLER (version 10.5, accessed on 30 November 2023) algorithm can be harnessed for this purpose since it computes intermediate models during simulated annealing, enabling the exploration of attainable configurational states and energies while minimizing its objective function. There have been a few attempts to date to improve the models generated by its algorithm, and in particular, there is no literature regarding the implementation of an averaging procedure involving the intermediate models in the MODELLER algorithm. In this study, we examined MODELLER's output using 225 target-template pairs, extracting the best representatives of intermediate models. Applying an averaging procedure to the selected intermediate structures based on statistical potentials, we aimed to determine: (1) whether averaging improves the quality of structural models during the building phase; (2) if ranking by statistical potentials reliably selects the best models, leading to improved final model quality; (3) whether using a single template versus multiple templates affects the averaging approach; (4) whether the "ensemble" nature of the MODELLER building phase can be harnessed to capture low-energy conformations in holo structures modeling. Our findings indicate that while improvements typically fall short of a few decimal points in the model evaluation metric, a notable fraction of configurations exhibit slightly higher similarity to the native structure than MODELLER's proposed final model. The averaging-building procedure proves particularly beneficial in (1) regions of low sequence identity between the target and template(s), the most challenging aspect of homology modeling; (2) holo protein conformations generation, an area in which MODELLER and related tools usually fall short of the expected performance.

Integration of pre-trained protein language models into geometric deep learning networks

Geometric deep learning has recently achieved great success in non-Euclidean domains, and learning on 3D structures of large biomolecules is emerging as a distinct research area. However, its efficacy is largely constrained due to the limited quantity of structural data. Meanwhile, protein language models trained on substantial 1D sequences have shown burgeoning capabilities with scale in a broad range of applications. Several preceding studies consider combining these different protein modalities to promote the representation power of geometric neural networks but fail to present a comprehensive understanding of their benefits. In this work, we integrate the knowledge learned by well-trained protein language models into several state-of-the-art geometric networks and evaluate a variety of protein representation learning benchmarks, including protein-protein interface prediction, model quality assessment, protein-protein rigid-body docking, and binding affinity prediction. Our findings show an overall improvement of 20% over baselines. Strong evidence indicates that the incorporation of protein language models' knowledge enhances geometric networks' capacity by a significant margin and can be generalized to complex tasks.

Contact-Assisted Threading in Low-Homology Protein Modeling

The ability to successfully predict the three-dimensional structure of a protein from its amino acid sequence has made considerable progress in the recent past. The progress is propelled by the improved accuracy of deep learning-based inter-residue contact map predictors coupled with the rising growth of protein sequence databases. Contact map encodes interatomic interaction information that can be exploited for highly accurate prediction of protein structures via contact map threading even for the query proteins that are not amenable to direct homology modeling. As such, contact-assisted threading has garnered considerable research effort. In this chapter, we provide an overview of existing contact-assisted threading methods while highlighting the recent advances and discussing some of the current limitations and future prospects in the application of contact-assisted threading for improving the accuracy of low-homology protein modeling.

3D-equivariant graph neural networks for protein model quality assessment

MOTIVATION

Quality assessment (QA) of predicted protein tertiary structure models plays an important role in ranking and using them. With the recent development of deep learning end-to-end protein structure prediction techniques for generating highly confident tertiary structures for most proteins, it is important to explore corresponding QA strategies to evaluate and select the structural models predicted by them since these models have better quality and different properties than the models predicted by traditional tertiary structure prediction methods.

RESULTS

We develop EnQA, a novel graph-based 3D-equivariant neural network method that is equivariant to rotation and translation of 3D objects to estimate the accuracy of protein structural models by leveraging the structural features acquired from the state-of-the-art tertiary structure prediction method-AlphaFold2. We train and test the method on both traditional model datasets (e.g. the datasets of the Critical Assessment of Techniques for Protein Structure Prediction) and a new dataset of high-quality structural models predicted only by AlphaFold2 for the proteins whose experimental structures were released recently. Our approach achieves state-of-the-art performance on protein structural models predicted by both traditional protein structure prediction methods and the latest end-to-end deep learning method-AlphaFold2. It performs even better than the model QA scores provided by AlphaFold2 itself. The results illustrate that the 3D-equivariant graph neural network is a promising approach to the evaluation of protein structural models. Integrating AlphaFold2 features with other complementary sequence and structural features is important for improving protein model QA.

AVAILABILITY AND IMPLEMENTATION

The source code is available at https://github.com/BioinfoMachineLearning/EnQA.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

DISTEMA: distance map-based estimation of single protein model accuracy with attentive 2D convolutional neural network

BACKGROUND

Estimation of the accuracy (quality) of protein structural models is important for both prediction and use of protein structural models. Deep learning methods have been used to integrate protein structure features to predict the quality of protein models. Inter-residue distances are key information for predicting protein's tertiary structures and therefore have good potentials to predict the quality of protein structural models. However, few methods have been developed to fully take advantage of predicted inter-residue distance maps to estimate the accuracy of a single protein structural model.

RESULT

We developed an attentive 2D convolutional neural network (CNN) with channel-wise attention to take only a raw difference map between the inter-residue distance map calculated from a single protein model and the distance map predicted from the protein sequence as input to predict the quality of the model. The network comprises multiple convolutional layers, batch normalization layers, dense layers, and Squeeze-and-Excitation blocks with attention to automatically extract features relevant to protein model quality from the raw input without using any expert-curated features. We evaluated DISTEMA's capability of selecting the best models for CASP13 targets in terms of ranking loss of GDT-TS score. The ranking loss of DISTEMA is 0.079, lower than several state-of-the-art single-model quality assessment methods.

CONCLUSION

This work demonstrates that using raw inter-residue distance information with deep learning can predict the quality of protein structural models reasonably well. DISTEMA is freely at https://github.com/jianlin-cheng/DISTEMA.

A Benchmark Dataset for Evaluating Practical Performance of Model Quality Assessment of Homology Models

Protein structure prediction is an important issue in structural bioinformatics. In this process, model quality assessment (MQA), which estimates the accuracy of the predicted structure, is also practically important. Currently, the most commonly used dataset to evaluate the performance of MQA is the critical assessment of the protein structure prediction (CASP) dataset. However, the CASP dataset does not contain enough targets with high-quality models, and thus cannot sufficiently evaluate the MQA performance in practical use. Additionally, most application studies employ homology modeling because of its reliability. However, the CASP dataset includes models generated by de novo methods, which may lead to the mis-estimation of MQA performance. In this study, we created new benchmark datasets, named a homology models dataset for model quality assessment (HMDM), that contain targets with high-quality models derived using homology modeling. We then benchmarked the performance of the MQA methods using the new datasets and compared their performance to that of the classical selection based on the sequence identity of the template proteins. The results showed that model selection by the latest MQA methods using deep learning is better than selection by template sequence identity and classical statistical potentials. Using HMDM, it is possible to verify the MQA performance for high-accuracy homology models.

ProFitFun: a protein tertiary structure fitness function for quantifying the accuracies of model structures

MOTIVATION

An accurate estimation of the quality of protein model structures typifies as a cornerstone in protein structure prediction regimes. Despite the recent groundbreaking success in the field of protein structure prediction, there are certain prospects for the improvement in model quality estimation at multiple stages of protein structure prediction and thus, to further push the prediction accuracy. Here, a novel approach, named ProFitFun, for assessing the quality of protein models is proposed by harnessing the sequence and structural features of experimental protein structures in terms of the preferences of backbone dihedral angles and relative surface accessibility of their amino acid residues at the tripeptide level. The proposed approach leverages upon the backbone dihedral angle and surface accessibility preferences of the residues by accounting for its N-terminal and C-terminal neighbors in the protein structure. These preferences are used to evaluate protein structures through a machine learning approach and tested on an extensive dataset of diverse proteins.

RESULTS

The approach was extensively validated on a large test dataset (n = 25 005) of protein structures, comprising 23 661 models of 82 non-homologous proteins and 1344 non-homologous experimental structures. In addition, an external dataset of 40 000 models of 200 non-homologous proteins was also used for the validation of the proposed method. Both datasets were further used for benchmarking the proposed method with four different state-of-the-art methods for protein structure quality assessment. In the benchmarking, the proposed method outperformed some state-of-the-art methods in terms of Spearman's and Pearson's correlation coefficients, average GDT-TS loss, sum of z-scores and average absolute difference of predictions over corresponding observed values. The high accuracy of the proposed approach promises a potential use of the sequence and structural features in computational protein design.

AVAILABILITY AND IMPLEMENTATION

http://github.com/KYZ-LSB/ProTerS-FitFun.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

MUfoldQA_G: High-accuracy protein model QA via retraining and transformation

Protein tertiary structure prediction is an active research area and has attracted significant attention recently due to the success of AlphaFold from DeepMind. Methods capable of accurately evaluating the quality of predicted models are of great importance. In the past, although many model quality assessment (QA) methods have been developed, their accuracies are not consistently high across different QA performance metrics for diverse target proteins. In this paper, we propose MUfoldQA_G, a new multi-model QA method that aims at simultaneously optimizing Pearson correlation and average GDT-TS difference, two commonly used QA performance metrics. This method is based on two new algorithms MUfoldQA_Gp and MUfoldQA_Gr. MUfoldQA_Gp uses a new technique to combine information from protein templates and reference protein models to maximize the Pearson correlation QA metric. MUfoldQA_Gr employs a new machine learning technique that resamples training data and retrains adaptively to learn a consensus model that is better than naïve consensus while minimizing average GDT-TS difference. MUfoldQA_G uses a new method to combine the results of MUfoldQA_Gr and MUfoldQA_Gp so that the final QA prediction results achieve low average GDT-TS difference that is close to the results from MUfoldQA_Gr, while maintaining high Pearson correlation that is the same as the results from MUfoldQA_Gp. In CASP14 QA categories, MUfoldQA_G ranked No. 1 in Pearson correlation and No. 2 in average GDT-TS difference.

Protein sequence-to-structure learning: Is this the end(-to-end revolution)?

The potential of deep learning has been recognized in the protein structure prediction community for some time, and became indisputable after CASP13. In CASP14, deep learning has boosted the field to unanticipated levels reaching near-experimental accuracy. This success comes from advances transferred from other machine learning areas, as well as methods specifically designed to deal with protein sequences and structures, and their abstractions. Novel emerging approaches include (i) geometric learning, that is, learning on representations such as graphs, three-dimensional (3D) Voronoi tessellations, and point clouds; (ii) pretrained protein language models leveraging attention; (iii) equivariant architectures preserving the symmetry of 3D space; (iv) use of large meta-genome databases; (v) combinations of protein representations; and (vi) finally truly end-to-end architectures, that is, differentiable models starting from a sequence and returning a 3D structure. Here, we provide an overview and our opinion of the novel deep learning approaches developed in the last 2 years and widely used in CASP14.

Improving protein tertiary structure prediction by deep learning and distance prediction in CASP14

Substantial progresses in protein structure prediction have been made by utilizing deep‐learning and residue‐residue distance prediction since CASP13. Inspired by the advances, we improve our CASP14 MULTICOM protein structure prediction system by incorporating three new components: (a) a new deep learning‐based protein inter‐residue distance predictor to improve template‐free (ab initio) tertiary structure prediction, (b) an enhanced template‐based tertiary structure prediction method, and (c) distance‐based model quality assessment methods empowered by deep learning. In the 2020 CASP14 experiment, MULTICOM predictor was ranked seventh out of 146 predictors in tertiary structure prediction and ranked third out of 136 predictors in inter‐domain structure prediction. The results demonstrate that the template‐free modeling based on deep learning and residue‐residue distance prediction can predict the correct topology for almost all template‐based modeling targets and a majority of hard targets (template‐free targets or targets whose templates cannot be recognized), which is a significant improvement over the CASP13 MULTICOM predictor. Moreover, the template‐free modeling performs better than the template‐based modeling on not only hard targets but also the targets that have homologous templates. The performance of the template‐free modeling largely depends on the accuracy of distance prediction closely related to the quality of multiple sequence alignments. The structural model quality assessment works well on targets for which enough good models can be predicted, but it may perform poorly when only a few good models are predicted for a hard target and the distribution of model quality scores is highly skewed. MULTICOM is available at https://github.com/jianlin-cheng/MULTICOM_Human_CASP14/tree/CASP14_DeepRank3 and https://github.com/multicom-toolbox/multicom/tree/multicom_v2.0.

Spherical convolutions on molecular graphs for protein model quality assessment

Processing information on three-dimensional (3D) objects requires methods stable to rigid-body transformations, in particular rotations, of the input data. In image processing tasks, convolutional neural networks achieve this property using rotation-equivariant operations. However, contrary to images, graphs generally have irregular topology. This makes it challenging to define a rotation-equivariant convolution operation on these structures. In this work, we propose spherical graph convolutional network that processes 3D models of proteins represented as molecular graphs. In a protein molecule, individual amino acids have common topological elements. This allows us to unambiguously associate each amino acid with a local coordinate system and construct rotation-equivariant spherical filters that operate on angular information between graph nodes. Within the framework of the protein model quality assessment problem, we demonstrate that the proposed spherical convolution method significantly improves the quality of model assessment compared to the standard message-passing approach. It is also comparable to state-of-the-art methods, as we demonstrate on critical assessment of structure prediction benchmarks. The proposed technique operates only on geometric features of protein 3D models. This makes it universal and applicable to any other geometric-learning task where the graph structure allows constructing local coordinate systems. The method is available at https://team.inria.fr/nano-d/software/s-gcn/.

MULTICOM2 open-source protein structure prediction system powered by deep learning and distance prediction

Protein structure prediction is an important problem in bioinformatics and has been studied for decades. However, there are still few open-source comprehensive protein structure prediction packages publicly available in the field. In this paper, we present our latest open-source protein tertiary structure prediction system—MULTICOM2, an integration of template-based modeling (TBM) and template-free modeling (FM) methods. The template-based modeling uses sequence alignment tools with deep multiple sequence alignments to search for structural templates, which are much faster and more accurate than MULTICOM1. The template-free (ab initio or de novo) modeling uses the inter-residue distances predicted by DeepDist to reconstruct tertiary structure models without using any known structure as template. In the blind CASP14 experiment, the average TM-score of the models predicted by our server predictor based on the MULTICOM2 system is 0.720 for 58 TBM (regular) domains and 0.514 for 38 FM and FM/TBM (hard) domains, indicating that MULTICOM2 is capable of predicting good tertiary structures across the board. It can predict the correct fold for 76 CASP14 domains (95% regular domains and 55% hard domains) if only one prediction is made for a domain. The success rate is increased to 3% for both regular and hard domains if five predictions are made per domain. Moreover, the prediction accuracy of the pure template-free structure modeling method on both TBM and FM targets is very close to the combination of template-based and template-free modeling methods. This demonstrates that the distance-based template-free modeling method powered by deep learning can largely replace the traditional template-based modeling method even on TBM targets that TBM methods used to dominate and therefore provides a uniform structure modeling approach to any protein. Finally, on the 38 CASP14 FM and FM/TBM hard domains, MULTICOM2 server predictors (MULTICOM-HYBRID, MULTICOM-DEEP, MULTICOM-DIST) were ranked among the top 20 automated server predictors in the CASP14 experiment. After combining multiple predictors from the same research group as one entry, MULTICOM-HYBRID was ranked no. 5. The source code of MULTICOM2 is freely available at https://github.com/multicom-toolbox/multicom/tree/multicom_v2.0.

KORP-PL: a coarse-grained knowledge-based scoring function for protein-ligand interactions

MOTIVATION

Despite the progress made in studying protein-ligand interactions and the widespread application of docking and affinity prediction tools, improving their precision and efficiency still remains a challenge. Computational approaches based on the scoring of docking conformations with statistical potentials constitute a popular alternative to more accurate but costly physics-based thermodynamic sampling methods. In this context, a minimalist and fast sidechain-free knowledge-based potential with a high docking and screening power can be very useful when screening a big number of putative docking conformations.

RESULTS

Here, we present a novel coarse-grained potential defined by a 3D joint probability distribution function that only depends on the pairwise orientation and position between protein backbone and ligand atoms. Despite its extreme simplicity, our approach yields very competitive results with the state-of-the-art scoring functions, especially in docking and screening tasks. For example, we observed a twofold improvement in the median 5% enrichment factor on the DUD-E benchmark compared to Autodock Vina results. Moreover, our results prove that a coarse sidechain-free potential is sufficient for a very successful docking pose prediction.

AVAILABILITYAND IMPLEMENTATION

The standalone version of KORP-PL with the corresponding tests and benchmarks are available at https://team.inria.fr/nano-d/korp-pl/ and https://chaconlab.org/modeling/korp-pl.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Protein model accuracy estimation empowered by deep learning and inter-residue distance prediction in CASP14

The inter-residue contact prediction and deep learning showed the promise to improve the estimation of protein model accuracy (EMA) in the 13th Critical Assessment of Protein Structure Prediction (CASP13). To further leverage the improved inter-residue distance predictions to enhance EMA, during the 2020 CASP14 experiment, we integrated several new inter-residue distance features with the existing model quality assessment features in several deep learning methods to predict the quality of protein structural models. According to the evaluation of performance in selecting the best model from the models of CASP14 targets, our three multi-model predictors of estimating model accuracy (MULTICOM-CONSTRUCT, MULTICOM-AI, and MULTICOM-CLUSTER) achieve the averaged loss of 0.073, 0.079, and 0.081, respectively, in terms of the global distance test score (GDT-TS). The three methods are ranked first, second, and third out of all 68 CASP14 predictors. MULTICOM-DEEP, the single-model predictor of estimating model accuracy (EMA), is ranked within top 10 among all the single-model EMA methods according to GDT-TS score loss. The results demonstrate that inter-residue distance features are valuable inputs for deep learning to predict the quality of protein structural models. However, larger training datasets and better ways of leveraging inter-residue distance information are needed to fully explore its potentials.

Recent Advances in Protein Homology Detection Propelled by Inter-Residue Interaction Map Threading

Sequence-based protein homology detection has emerged as one of the most sensitive and accurate approaches to protein structure prediction. Despite the success, homology detection remains very challenging for weakly homologous proteins with divergent evolutionary profile. Very recently, deep neural network architectures have shown promising progress in mining the coevolutionary signal encoded in multiple sequence alignments, leading to reasonably accurate estimation of inter-residue interaction maps, which serve as a rich source of additional information for improved homology detection. Here, we summarize the latest developments in protein homology detection driven by inter-residue interaction map threading. We highlight the emerging trends in distant-homology protein threading through the alignment of predicted interaction maps at various granularities ranging from binary contact maps to finer-grained distance and orientation maps as well as their combination. We also discuss some of the current limitations and possible future avenues to further enhance the sensitivity of protein homology detection.

GraphQA: protein model quality assessment using graph convolutional networks

Motivation

Proteins are ubiquitous molecules whose function in biological processes is determined by their 3D structure. Experimental identification of a protein’s structure can be time-consuming, prohibitively expensive and not always possible. Alternatively, protein folding can be modeled using computational methods, which however are not guaranteed to always produce optimal results. GraphQA is a graph-based method to estimate the quality of protein models, that possesses favorable properties such as representation learning, explicit modeling of both sequential and 3D structure, geometric invariance and computational efficiency.

Results

GraphQA performs similarly to state-of-the-art methods despite using a relatively low number of input features. In addition, the graph network structure provides an improvement over the architecture used in ProQ4 operating on the same input features. Finally, the individual contributions of GraphQA components are carefully evaluated.

Availability and implementation

PyTorch implementation, datasets, experiments and link to an evaluation server are available through this GitHub repository: github.com/baldassarreFe/graphqa.

Supplementary information

Supplementary data are available at Bioinformatics online.

P3CMQA: Single-Model Quality Assessment Using 3DCNN with Profile-Based Features

Model quality assessment (MQA), which selects near-native structures from structure models, is an important process in protein tertiary structure prediction. The three-dimensional convolution neural network (3DCNN) was applied to the task, but the performance was comparable to existing methods because it used only atom-type features as the input. Thus, we added sequence profile-based features, which are also used in other methods, to improve the performance. We developed a single-model MQA method for protein structures based on 3DCNN using sequence profile-based features, namely, P3CMQA. Performance evaluation using a CASP13 dataset showed that profile-based features improved the assessment performance, and the proposed method was better than currently available single-model MQA methods, including the previous 3DCNN-based method. We also implemented a web-interface of the method to make it more user-friendly.

Geometric potentials from deep learning improve prediction of CDR H3 loop structures

Motivation

Antibody structure is largely conserved, except for a complementarity-determining region featuring six variable loops. Five of these loops adopt canonical folds which can typically be predicted with existing methods, while the remaining loop (CDR H3) remains a challenge due to its highly diverse set of observed conformations. In recent years, deep neural networks have proven to be effective at capturing the complex patterns of protein structure. This work proposes DeepH3, a deep residual neural network that learns to predict inter-residue distances and orientations from antibody heavy and light chain sequence. The output of DeepH3 is a set of probability distributions over distances and orientation angles between pairs of residues. These distributions are converted to geometric potentials and used to discriminate between decoy structures produced by RosettaAntibody and predict new CDR H3 loop structures de novo.

Results

When evaluated on the Rosetta antibody benchmark dataset of 49 targets, DeepH3-predicted potentials identified better, same and worse structures [measured by root-mean-squared distance (RMSD) from the experimental CDR H3 loop structure] than the standard Rosetta energy function for 33, 6 and 10 targets, respectively, and improved the average RMSD of predictions by 32.1% (1.4 Å). Analysis of individual geometric potentials revealed that inter-residue orientations were more effective than inter-residue distances for discriminating near-native CDR H3 loops. When applied to de novo prediction of CDR H3 loop structures, DeepH3 achieves an average RMSD of 2.2 ± 1.1 Å on the Rosetta antibody benchmark.

Availability and Implementation

DeepH3 source code and pre-trained model parameters are freely available at https://github.com/Graylab/deepH3-distances-orientations.

Supplementary information

Supplementary data are available at Bioinformatics online.

QDeep: distance-based protein model quality estimation by residue-level ensemble error classifications using stacked deep residual neural networks

MOTIVATION

Protein model quality estimation, in many ways, informs protein structure prediction. Despite their tight coupling, existing model quality estimation methods do not leverage inter-residue distance information or the latest technological breakthrough in deep learning that has recently revolutionized protein structure prediction.

RESULTS

We present a new distance-based single-model quality estimation method called QDeep by harnessing the power of stacked deep residual neural networks (ResNets). Our method first employs stacked deep ResNets to perform residue-level ensemble error classifications at multiple predefined error thresholds, and then combines the predictions from the individual error classifiers for estimating the quality of a protein structural model. Experimental results show that our method consistently outperforms existing state-of-the-art methods including ProQ2, ProQ3, ProQ3D, ProQ4, 3DCNN, MESHI, and VoroMQA in multiple independent test datasets across a wide-range of accuracy measures; and that predicted distance information significantly contributes to the improved performance of QDeep.

AVAILABILITY AND IMPLEMENTATION

https://github.com/Bhattacharya-Lab/QDeep.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

VoroCNN: Deep convolutional neural network built on 3D Voronoi tessellation of protein structures

MOTIVATION

Effective use of evolutionary information has recently led to tremendous progress in computational prediction of three-dimensional (3D) structures of proteins and their complexes. Despite the progress, the accuracy of predicted structures tends to vary considerably from case to case. Since the utility of computational models depends on their accuracy, reliable estimates of deviation between predicted and native structures are of utmost importance.

RESULTS

For the first time, we present a deep convolutional neural network (CNN) constructed on a Voronoi tessellation of 3D molecular structures. Despite the irregular data domain, our data representation allows us to efficiently introduce both convolution and pooling operations and train the network in an end-to-end fashion without precomputed descriptors. The resultant model, VoroCNN, predicts local qualities of 3D protein folds. The prediction results are competitive to state of the art and superior to the previous 3D CNN architectures built for the same task. We also discuss practical applications of VoroCNN, for example, in recognition of protein binding interfaces.

AVAILABILITY

The model, data, and evaluation tests are available at https://team.inria.fr/nano-d/software/vorocnn/.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Unsupervised multi-instance learning for protein structure determination

Many regions of the protein universe remain inaccessible by wet-laboratory or computational structure determination methods. A significant challenge in elucidating these dark regions in silico relates to the ability to discriminate relevant structure(s) among many structures/decoys computed for a protein of interest, a problem known as decoy selection. Clustering decoys based on geometric similarity remains popular. However, it is unclear how exactly to exploit the groups of decoys revealed via clustering to select individual structures for prediction. In this paper, we provide an intuitive formulation of the decoy selection problem as an instance of unsupervised multi-instance learning. We address the problem in three stages, first organizing given decoys of a protein molecule into bags, then identifying relevant bags, and finally drawing individual instances from these bags to offer as prediction. We propose both non-parametric and parametric algorithms for drawing individual instances. Our evaluation utilizes two datasets, one benchmark dataset of ensembles of decoys for a varied list of protein molecules, and a dataset of decoy ensembles for targets drawn from recent CASP competitions. A comparative analysis with state-of-the-art methods reveals that the proposed approach outperforms existing methods, thus warranting further investigation of multi-instance learning to advance our treatment of decoy selection.

DeepDist: real-value inter-residue distance prediction with deep residual convolutional network

BACKGROUND

Driven by deep learning, inter-residue contact/distance prediction has been significantly improved and substantially enhanced ab initio protein structure prediction. Currently, most of the distance prediction methods classify inter-residue distances into multiple distance intervals instead of directly predicting real-value distances. The output of the former has to be converted into real-value distances to be used in tertiary structure prediction.

RESULTS

To explore the potentials of predicting real-value inter-residue distances, we develop a multi-task deep learning distance predictor (DeepDist) based on new residual convolutional network architectures to simultaneously predict real-value inter-residue distances and classify them into multiple distance intervals. Tested on 43 CASP13 hard domains, DeepDist achieves comparable performance in real-value distance prediction and multi-class distance prediction. The average mean square error (MSE) of DeepDist's real-value distance prediction is 0.896 Ų when filtering out the predicted distance ≥ 16 Å, which is lower than 1.003 Ų of DeepDist's multi-class distance prediction. When distance predictions are converted into contact predictions at 8 Å threshold (the standard threshold in the field), the precision of top L/5 and L/2 contact predictions of DeepDist's multi-class distance prediction is 79.3% and 66.1%, respectively, higher than 78.6% and 64.5% of its real-value distance prediction and the best results in the CASP13 experiment.

CONCLUSIONS

DeepDist can predict inter-residue distances well and improve binary contact prediction over the existing state-of-the-art methods. Moreover, the predicted real-value distances can be directly used to reconstruct protein tertiary structures better than multi-class distance predictions due to the lower MSE. Finally, we demonstrate that predicting the real-value distance map and multi-class distance map at the same time performs better than predicting real-value distances alone.

Improved Protein Model Quality Assessment By Integrating Sequential And Pairwise Features Using Deep Learning

MOTIVATION

Accurately estimating protein model quality in the absence of experimental structure is not only important for model evaluation and selection, but also useful for model refinement. Progress has been steadily made by introducing new features and algorithms (especially deep neural networks), but the accuracy of quality assessment (QA) is still not very satisfactory, especially local QA on hard protein targets.

RESULTS

We propose a new single-model-based QA method ResNetQA for both local and global quality assessment. Our method predicts model quality by integrating sequential and pairwise features using a deep neural network composed of both 1 D and 2 D convolutional residual neural networks (ResNet). The 2 D ResNet module extracts useful information from pairwise features such as model-derived distance maps, co-evolution information, and predicted distance potential from sequences. The 1 D ResNet is used to predict local (global) model quality from sequential features and pooled pairwise information generated by 2 D ResNet. Tested on the CASP12 and CASP13 datasets, our experimental results show that our method greatly outperforms existing state-of-the-art methods. Our ablation studies indicate that the 2 D ResNet module and pairwise features play an important role in improving model quality assessment.

AVAILABILITY

https://github.com/AndersJing/ResNetQA.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Homology modeling in the time of collective and artificial intelligence

Homology modeling is a method for building protein 3D structures using protein primary sequence and utilizing prior knowledge gained from structural similarities with other proteins. The homology modeling process is done in sequential steps where sequence/structure alignment is optimized, then a backbone is built and later, side-chains are added. Once the low-homology loops are modeled, the whole 3D structure is optimized and validated. In the past three decades, a few collective and collaborative initiatives allowed for continuous progress in both homology and ab initio modeling. Critical Assessment of protein Structure Prediction (CASP) is a worldwide community experiment that has historically recorded the progress in this field. Folding@Home and Rosetta@Home are examples of crowd-sourcing initiatives where the community is sharing computational resources, whereas RosettaCommons is an example of an initiative where a community is sharing a codebase for the development of computational algorithms. Foldit is another initiative where participants compete with each other in a protein folding video game to predict 3D structure. In the past few years, contact maps deep machine learning was introduced to the 3D structure prediction process, adding more information and increasing the accuracy of models significantly. In this review, we will take the reader in a journey of exploration from the beginnings to the most recent turnabouts, which have revolutionized the field of homology modeling. Moreover, we discuss the new trends emerging in this rapidly growing field.

Protein loops with multiple meta-stable conformations: A challenge for sampling and scoring methods

Flexible regions in proteins, such as loops, cannot be represented by a single conformation. Instead, conformational ensembles are needed to provide a more global picture. In this context, identifying statistically meaningful conformations within an ensemble generated by loop sampling techniques remains an open problem. The difficulty is primarily related to the lack of structural data about these flexible regions. With the majority of structural data coming from x-ray crystallography and ignoring plasticity, the conception and evaluation of loop scoring methods is challenging. In this work, we compare the performance of various scoring methods on a set of eight protein loops that are known to be flexible. The ability of each method to identify and select all of the known conformations is assessed, and the underlying energy landscapes are produced and projected to visualize the qualitative differences obtained when using the methods. Statistical potentials are found to provide considerable reliability despite their being designed to tradeoff accuracy for lower computational cost. On a large pool of loop models, they are capable of filtering out statistically improbable states while retaining those that resemble known (and thus likely) conformations. However, computationally expensive methods are still required for more precise assessment and structural refinement. The results also highlight the importance of employing several scaffolds for the protein, due to the high influence of small structural rearrangements in the rest of the protein over the modeled energy landscape for the loop.

Using physical features of protein core packing to distinguish real proteins from decoys

The ability to consistently distinguish real protein structures from computationally generated model decoys is not yet a solved problem. One route to distinguish real protein structures from decoys is to delineate the important physical features that specify a real protein. For example, it has long been appreciated that the hydrophobic cores of proteins contribute significantly to their stability. We used two sources to obtain datasets of decoys to compare with real protein structures: submissions to the biennial Critical Assessment of protein Structure Prediction competition, in which researchers attempt to predict the structure of a protein only knowing its amino acid sequence, and also decoys generated by 3DRobot, which have user-specified global root-mean-squared deviations from experimentally determined structures. Our analysis revealed that both sets of decoys possess cores that do not recapitulate the key features that define real protein cores. In particular, the model structures appear more densely packed (because of energetically unfavorable atomic overlaps), contain too few residues in the core, and have improper distributions of hydrophobic residues throughout the structure. Based on these observations, we developed a feed-forward neural network, which incorporates key physical features of protein cores, to predict how well a computational model recapitulates the real protein structure without knowledge of the structure of the target sequence. By identifying the important features of protein structure, our method is able to rank decoy structures with similar accuracy to that obtained by state-of-the-art methods that incorporate many additional features. The small number of physical features makes our model interpretable, emphasizing the importance of protein packing and hydrophobicity in protein structure prediction.

VoroMQA web server for assessing three-dimensional structures of proteins and protein complexes

The VoroMQA (Voronoi tessellation-based Model Quality Assessment) web server is dedicated to the estimation of protein structure quality, a common step in selecting realistic and most accurate computational models and in validating experimental structures. As an input, the VoroMQA web server accepts one or more protein structures in PDB format. Input structures may be either monomeric proteins or multimeric protein complexes. For every input structure, the server provides both global and local (per-residue) scores. Visualization of the local scores along the protein chain is enhanced by providing secondary structure assignment and information on solvent accessibility. A unique feature of the VoroMQA server is the ability to directly assess protein-protein interaction interfaces. If this type of assessment is requested, the web server provides interface quality scores, interface energy estimates, and local scores for residues involved in inter-chain interfaces. VoroMQA, the underlying method of the web server, was extensively tested in recent community-wide CASP and CAPRI experiments. During these experiments VoroMQA showed outstanding performance both in model selection and in estimation of accuracy of local structural regions. The VoroMQA web server is available at http://bioinformatics.ibt.lt/wtsam/voromqa.

Neighborhood Preference of Amino Acids in Protein Structures and its Applications in Protein Structure Assessment

Amino acids form protein 3D structures in unique manners such that the folded structure is stable and functional under physiological conditions. Non-specific and non-covalent interactions between amino acids exhibit neighborhood preferences. Based on structural information from the protein data bank, a statistical energy function was derived to quantify amino acid neighborhood preferences. The neighborhood of one amino acid is defined by its contacting residues, and the energy function is determined by the neighboring residue types and relative positions. The neighborhood preference of amino acids was exploited to facilitate structural quality assessment, which was implemented in the neighborhood preference program NEPRE. The source codes are available via https://github.com/LiuLab-CSRC/NePre.

Improved Consensus-Fragment Selection in Template-Assisted Prediction of Protein Structures with the UNRES Force Field in CASP13

The method for protein-structure prediction, which combines the physics-based coarse-grained UNRES force field with knowledge-based modeling, has been developed further and tested in the 13th Community Wide Experiment on the Critical Assessment of Techniques for Protein Structure Prediction (CASP13). The method implements restraints from the consensus fragments common to server models. In this work, the server models to derive fragments have been chosen on the basis of quality assessment; a fully automatic fragment-selection procedure has been introduced, and Dynamic Fragment Assembly pseudopotentials have been fully implemented. The Global Distance Test Score (GDT_TS), averaged over our “Model 1” predictions, increased by over 10 units with respect to CASP12 for the free-modeling category to reach 40.82. Our “Model 1” predictions ranked 20 and 14 for all and free-modeling targets, respectively (upper 20.2% and 14.3% of all models submitted to CASP13 in these categories, respectively), compared to 27 (upper 21.1%) and 24 (upper 18.9%) in CASP12, respectively. For oligomeric targets, the Interface Patch Similarity (IPS) and Interface Contact Similarity (ICS) averaged over our best oligomer models increased from 0.28 to 0.36 and from 12.4 to 17.8, respectively, from CASP12 to CASP13, and top-ranking models of 2 targets (H0968 and T0997o) were obtained (none in CASP12). The improvement of our method in CASP13 over CASP12 was ascribed to the combined effect of the overall enhancement of server-model quality, our success in selecting server models and fragments to derive restraints, and improvements of the restraint and potential-energy functions.

Assessment of chemical-crosslink-assisted protein structure modeling in CASP13

With the advance of experimental procedures obtaining chemical crosslinking information is becoming a fast and routine practice. Information on crosslinks can greatly enhance the accuracy of protein structure modeling. Here, we review the current state of the art in modeling protein structures with the assistance of experimentally determined chemical crosslinks within the framework of the 13th meeting of Critical Assessment of Structure Prediction approaches. This largest-to-date blind assessment reveals benefits of using data assistance in difficult to model protein structure prediction cases. However, in a broader context, it also suggests that with the unprecedented advance in accuracy to predict contacts in recent years, experimental crosslinks will be useful only if their specificity and accuracy further improved and they are better integrated into computational workflows.

Unsupervised and Supervised Learning over theEnergy Landscape for Protein Decoy Selection

The energy landscape that organizes microstates of a molecular system and governs theunderlying molecular dynamics exposes the relationship between molecular form/structure, changesto form, and biological activity or function in the cell. However, several challenges stand in the wayof leveraging energy landscapes for relating structure and structural dynamics to function. Energylandscapes are high-dimensional, multi-modal, and often overly-rugged. Deep wells or basins inthem do not always correspond to stable structural states but are instead the result of inherentinaccuracies in semi-empirical molecular energy functions. Due to these challenges, energeticsis typically ignored in computational approaches addressing long-standing central questions incomputational biology, such as protein decoy selection. In the latter, the goal is to determine over apossibly large number of computationally-generated three-dimensional structures of a protein thosestructures that are biologically-active/native. In recent work, we have recast our attention on theprotein energy landscape and its role in helping us to advance decoy selection. Here, we summarizesome of our successes so far in this direction via unsupervised learning. More importantly, we furtheradvance the argument that the energy landscape holds valuable information to aid and advance thestate of protein decoy selection via novel machine learning methodologies that leverage supervisedlearning. Our focus in this article is on decoy selection for the purpose of a rigorous, quantitativeevaluation of how leveraging protein energy landscapes advances an important problem in proteinmodeling. However, the ideas and concepts presented here are generally useful to make discoveriesin studies aiming to relate molecular structure and structural dynamics to function.

Estimation of model accuracy in CASP13

Methods to reliably estimate the accuracy of 3D models of proteins are both a fundamental part of most protein folding pipelines and important for reliable identification of the best models when multiple pipelines are used. Here, we describe the progress made from CASP12 to CASP13 in the field of estimation of model accuracy (EMA) as seen from the progress of the most successful methods in CASP13. We show small but clear progress, that is, several methods perform better than the best methods from CASP12 when tested on CASP13 EMA targets. Some progress is driven by applying deep learning and residue-residue contacts to model accuracy prediction. We show that the best EMA methods select better models than the best servers in CASP13, but that there exists a great potential to improve this further. Also, according to the evaluation criteria based on local similarities, such as lDDT and CAD, it is now clear that single model accuracy methods perform relatively better than consensus-based methods.

ANDIS: an atomic angle- and distance-dependent statistical potential for protein structure quality assessment

Background

The knowledge-based statistical potential has been widely used in protein structure modeling and model quality assessment. They are commonly evaluated based on their abilities of native recognition as well as decoy discrimination. However, these two aspects are found to be mutually exclusive in many statistical potentials.

Results

We developed an atomic ANgle- and DIStance-dependent (ANDIS) statistical potential for protein structure quality assessment with distance cutoff being a tunable parameter. When distance cutoff is ≤9.0 Å, “effective atomic interaction” is employed to enhance the ability of native recognition. For a distance cutoff of ≥10 Å, the distance-dependent atom-pair potential with random-walk reference state is combined to strengthen the ability of decoy discrimination. Benchmark tests on 632 structural decoy sets from diverse sources demonstrate that ANDIS outperforms other state-of-the-art potentials in both native recognition and decoy discrimination.

Conclusions

Distance cutoff is a crucial parameter for distance-dependent statistical potentials. A lower distance cutoff is better for native recognition, while a higher one is favorable for decoy discrimination. The ANDIS potential is freely available as a standalone application at http://qbp.hzau.edu.cn/ANDIS/.

Electronic supplementary material

The online version of this article (10.1186/s12859-019-2898-y) contains supplementary material, which is available to authorized users.

Protein tertiary structure modeling driven by deep learning and contact distance prediction in CASP13

Predicting residue‐residue distance relationships (eg, contacts) has become the key direction to advance protein structure prediction since 2014 CASP11 experiment, while deep learning has revolutionized the technology for contact and distance distribution prediction since its debut in 2012 CASP10 experiment. During 2018 CASP13 experiment, we enhanced our MULTICOM protein structure prediction system with three major components: contact distance prediction based on deep convolutional neural networks, distance‐driven template‐free (ab initio) modeling, and protein model ranking empowered by deep learning and contact prediction. Our experiment demonstrates that contact distance prediction and deep learning methods are the key reasons that MULTICOM was ranked 3rd out of all 98 predictors in both template‐free and template‐based structure modeling in CASP13. Deep convolutional neural network can utilize global information in pairwise residue‐residue features such as coevolution scores to substantially improve contact distance prediction, which played a decisive role in correctly folding some free modeling and hard template‐based modeling targets. Deep learning also successfully integrated one‐dimensional structural features, two‐dimensional contact information, and three‐dimensional structural quality scores to improve protein model quality assessment, where the contact prediction was demonstrated to consistently enhance ranking of protein models for the first time. The success of MULTICOM system clearly shows that protein contact distance prediction and model selection driven by deep learning holds the key of solving protein structure prediction problem. However, there are still challenges in accurately predicting protein contact distance when there are few homologous sequences, folding proteins from noisy contact distances, and ranking models of hard targets.

Protein model quality assessment using 3D oriented convolutional neural networks

Motivation

Protein model quality assessment (QA) is a crucial and yet open problem in structural bioinformatics. The current best methods for single-model QA typically combine results from different approaches, each based on different input features constructed by experts in the field. Then, the prediction model is trained using a machine-learning algorithm. Recently, with the development of convolutional neural networks (CNN), the training paradigm has changed. In computer vision, the expert-developed features have been significantly overpassed by automatically trained convolutional filters. This motivated us to apply a three-dimensional (3D) CNN to the problem of protein model QA.

Results

We developed Ornate (Oriented Routed Neural network with Automatic Typing)—a novel method for single-model QA. Ornate is a residue-wise scoring function that takes as input 3D density maps. It predicts the local (residue-wise) and the global model quality through a deep 3D CNN. Specifically, Ornate aligns the input density map, corresponding to each residue and its neighborhood, with the backbone topology of this residue. This circumvents the problem of ambiguous orientations of the initial models. Also, Ornate includes automatic identification of atom types and dynamic routing of the data in the network. Established benchmarks (CASP 11 and CASP 12) demonstrate the state-of-the-art performance of our approach among single-model QA methods.

Availability and implementation

The method is available at https://team.inria.fr/nano-d/software/Ornate/. It consists of a C++ executable that transforms molecular structures into volumetric density maps, and a Python code based on the TensorFlow framework for applying the Ornate model to these maps.

Supplementary information

Supplementary data are available at Bioinformatics online.