GOAP: a generalized orientation-dependent, all-atom statistical potential for protein structure prediction

Contriving a novel of CHB therapeutic vaccine based on IgV_CTLA-4 and L protein via immunoinformatics approach

Chronic infection induced by immune tolerance to hepatitis B virus (HBV) is one of the most common causes of hepatic cirrhosis and hepatoma. Fortunately, the application of therapeutic vaccine can not only reverse HBV-tolerance, but also serve a potentially effective therapeutic strategy for treating chronic hepatitis B (CHB). However, the clinical effect of the currently developed CHB therapeutic vaccine is not optimistic due to the weak immunogenicity. Given that the human leukocyte antigen CTLA-4 owns strong binding ability to the surface B7 molecules (CD80 and CD86) of antigen presenting cell (APCs), the immunoglobulin variable region of CTLA-4 (IgV_CTLA-4) was fused with the L protein of HBV to contrive a novel therapeutic vaccine (V_C4HBL) for CHB in this study. We found that the addition of IgV_CTLA-4 did not interfere with the formation of L protein T cell and B cell epitopes after analysis by means of immunoinformatics approaches. Meanwhile, we also found that the IgV_CTLA-4 had strong binding force to B7 molecules through molecular docking and molecular dynamics (MD) simulation. Notably, our vaccine V_C4HBL showed good immunogenicity and antigenicity by in vitro and in vivo experiments. Therefore, the V_C4HBL is promising to again effectively activate the cellular and humoral immunity of CHB patients, and provides a potentially effective therapeutic strategy for the treatment of CHB in the future.Communicated by Ramaswamy H. Sarma.

SurfPro-NN: A 3D point cloud neural network for the scoring of protein-protein docking models based on surfaces features and protein language models

Protein-protein interactions (PPI) play a crucial role in numerous key biological processes, and the structure of protein complexes provides valuable clues for in-depth exploration of molecular-level biological processes. Protein-protein docking technology is widely used to simulate the spatial structure of proteins. However, there are still challenges in selecting candidate decoys that closely resemble the native structure from protein-protein docking simulations. In this study, we introduce a docking evaluation method based on three-dimensional point cloud neural networks named SurfPro-NN, which represents protein structures as point clouds and learns interaction information from protein interfaces by applying a point cloud neural network. With the continuous advancement of deep learning in the field of biology, a series of knowledge-rich pre-trained models have emerged. We incorporate protein surface representation models and language models into our approach, greatly enhancing feature representation capabilities and achieving superior performance in protein docking model scoring tasks. Through comprehensive testing on public datasets, we find that our method outperforms state-of-the-art deep learning approaches in protein-protein docking model scoring. Not only does it significantly improve performance, but it also greatly accelerates training speed. This study demonstrates the potential of our approach in addressing protein interaction assessment problems, providing strong support for future research and applications in the field of biology.

Atom-ProteinQA: Atom-level protein model quality assessment through fine-grained joint learning

MOTIVATION

Protein model quality assessment (ProteinQA) is a fundamental task that is essential for biologically relevant applications, i.e., protein structure refinement, protein design, etc. Previous works aimed to conduct ProteinQA only on the global structure or per-residue level, ignoring potentially usable and precise cues from a fine-grained per-atom perspective. In this study, we propose an atom-level ProteinQA model, named Atom-ProteinQA, in which two innovative modules are designed to extract geometric and topological atom-level relationships respectively. Specifically, on the one hand, a geometric perception module exploits 3D sparse convolution to capture the geometric features of the input protein, generating fine-grained atom-level predictions. On the other hand, natural chemical bonds are utilized to construct an atom-level graph, then message passing from a topological perception module is applied to output residue-level predictions in parallel. Eventually, through a cross-model aggregation module, features from different modules mutually interact, enhancing performance on both the atom and residue levels.

RESULTS

Extensive experiments show that our proposed Atom-ProteinQA outperforms previous methods by a large margin, regardless of residue-level or atom-level assessment. Concretely, we achieved state-of-the-art performance on CATH-2084, Decoy-8000, public benchmarks CASP13 & CASP14, and the CAMEO.

AVAILABILITY

The repository of this project is released on: https://github.com/luyfcandy/Atom_ProteinQA.

A Survey of Deep Learning Methods for Estimating the Accuracy of Protein Quaternary Structure Models

The quality prediction of quaternary structure models of a protein complex, in the absence of its true structure, is known as the Estimation of Model Accuracy (EMA). EMA is useful for ranking predicted protein complex structures and using them appropriately in biomedical research, such as protein-protein interaction studies, protein design, and drug discovery. With the advent of more accurate protein complex (multimer) prediction tools, such as AlphaFold2-Multimer and ESMFold, the estimation of the accuracy of protein complex structures has attracted increasing attention. Many deep learning methods have been developed to tackle this problem; however, there is a noticeable absence of a comprehensive overview of these methods to facilitate future development. Addressing this gap, we present a review of deep learning EMA methods for protein complex structures developed in the past several years, analyzing their methodologies, data and feature construction. We also provide a prospective summary of some potential new developments for further improving the accuracy of the EMA methods.

Assembly of Protein Complexes in and on the Membrane with Predicted Spatial Arrangement Constraints

Membrane proteins play crucial roles in various cellular processes, and their interactions with other proteins in and on the membrane are essential for their proper functioning. While an increasing number of structures of more membrane proteins are being determined, the available structure data is still sparse. To gain insights into the mechanisms of membrane protein complexes, computational docking methods are necessary due to the challenge of experimental determination. Here, we introduce Mem-LZerD, a rigid-body membrane docking algorithm designed to take advantage of modern membrane modeling and protein docking techniques to facilitate the docking of membrane protein complexes. Mem-LZerD is based on the LZerD protein docking algorithm, which has been constantly among the top servers in many rounds of CAPRI protein docking assessment. By employing a combination of geometric hashing, newly constrained by the predicted membrane height and tilt angle, and model scoring accounting for the energy of membrane insertion, we demonstrate the capability of Mem-LZerD to model diverse membrane protein-protein complexes. Mem-LZerD successfully performed unbound docking on 13 of 21 (61.9%) transmembrane complexes in an established benchmark, more than shown by previous approaches. It was additionally tested on new datasets of 44 transmembrane complexes and 92 peripheral membrane protein complexes, of which it successfully modeled 35 (79.5%) and 15 (16.3%) complexes respectively. When non-blind orientations of peripheral targets were included, the number of successes increased to 54 (58.7%). We further demonstrate that Mem-LZerD produces complex models which are suitable for molecular dynamics simulation. Mem-LZerD is made available at https://lzerd.kiharalab.org.

DeepRank-GNN-esm: a graph neural network for scoring protein-protein models using protein language model

Motivation

Protein-Protein interactions (PPIs) play critical roles in numerous cellular processes. By modelling the 3D structures of the correspond protein complexes valuable insights can be obtained, providing, e.g. starting points for drug and protein design. One challenge in the modelling process is however the identification of near-native models from the large pool of generated models. To this end we have previously developed DeepRank-GNN, a graph neural network that integrates structural and sequence information to enable effective pattern learning at PPI interfaces. Its main features are related to the Position Specific Scoring Matrices (PSSMs), which are computationally expensive to generate, significantly limits the algorithm's usability.

Results

We introduce here DeepRank-GNN-esm that includes as additional features protein language model embeddings from the ESM-2 model. We show that the ESM-2 embeddings can actually replace the PSSM features at no cost in-, or even better performance on two PPI-related tasks: scoring docking poses and detecting crystal artifacts. This new DeepRank version bypasses thus the need of generating PSSM, greatly improving the usability of the software and opening new application opportunities for systems for which PSSM profiles cannot be obtained or are irrelevant (e.g. antibody-antigen complexes).

Availability and implementation

DeepRank-GNN-esm is freely available from https://github.com/DeepRank/DeepRank-GNN-esm.

Assessment of Protein-Protein Docking Models Using Deep Learning

Protein-protein interactions are involved in almost all processes in a living cell and determine the biological functions of proteins. To obtain mechanistic understandings of protein-protein interactions, the tertiary structures of protein complexes have been determined by biophysical experimental methods, such as X-ray crystallography and cryogenic electron microscopy. However, as experimental methods are costly in resources, many computational methods have been developed that model protein complex structures. One of the difficulties in computational protein complex modeling (protein docking) is to select the most accurate models among many models that are usually generated by a docking method. This article reviews advances in protein docking model assessment methods, focusing on recent developments that apply deep learning to several network architectures.

Assembly of Protein Complexes In and On the Membrane with Predicted Spatial Arrangement Constraints

Membrane proteins play crucial roles in various cellular processes, and their interactions with other proteins in and on the membrane are essential for their proper functioning. While an increasing number of structures of more membrane proteins are being determined, the available structure data is still sparse. To gain insights into the mechanisms of membrane protein complexes, computational docking methods are necessary due to the challenge of experimental determination. Here, we introduce Mem-LZerD, a rigid-body membrane docking algorithm designed to take advantage of modern membrane modeling and protein docking techniques to facilitate the docking of membrane protein complexes. Mem-LZerD is based on the LZerD protein docking algorithm, which has been constantly among the top servers in many rounds of CAPRI protein docking assessment. By employing a combination of geometric hashing, newly constrained by the predicted membrane height and tilt angle, and model scoring accounting for the energy of membrane insertion, we demonstrate the capability of Mem-LZerD to model diverse membrane protein-protein complexes. Mem-LZerD successfully performed unbound docking on 13 of 21 (61.9%) transmembrane complexes in an established benchmark, more than shown by previous approaches. It was additionally tested on new datasets of 44 transmembrane complexes and 92 peripheral membrane protein complexes, of which it successfully modeled 35 (79.5%) and 15 (16.3%) complexes respectively. When non-blind orientations of peripheral targets were included, the number of successes increased to 54 (58.7%). We further demonstrate that Mem-LZerD produces complex models which are suitable for molecular dynamics simulation. Mem-LZerD is made available at https://lzerd.kiharalab.org.

The Molecular Docking of MAX Fungal Effectors with Plant HMA Domain-Binding Proteins

Fungal effector proteins are important in mediating disease infections in agriculturally important crops. These secreted small proteins are known to interact with their respective host receptor binding partners in the host, either inside the cells or in the apoplastic space, depending on the localisation of the effector proteins. Consequently, it is important to understand the interactions between fungal effector proteins and their target host receptor binding partners, particularly since this can be used for the selection of potential plant resistance or susceptibility-related proteins that can be applied to the breeding of new cultivars with disease resistance. In this study, molecular docking simulations were used to characterise protein-protein interactions between effector and plant receptors. Benchmarking was undertaken using available experimental structures of effector-host receptor complexes to optimise simulation parameters, which were then used to predict the structures and mediating interactions of effector proteins with host receptor binding partners that have not yet been characterised experimentally. Rigid docking was applied for both the so-called bound and unbound docking of MAX effectors with plant HMA domain protein partners. All bound complexes used for benchmarking were correctly predicted, with 84% being ranked as the top docking pose using the ZDOCK scoring function. In the case of unbound complexes, a minimum of 95% of known residues were predicted to be part of the interacting interface on the host receptor binding partner, and at least 87% of known residues were predicted to be part of the interacting interface on the effector protein. Hydrophobic interactions were found to dominate the formation of effector-plant protein complexes. An optimised set of docking parameters based on the use of ZDOCK and ZRANK scoring functions were established to enable the prediction of near-native docking poses involving different binding interfaces on plant HMA domain proteins. Whilst this study was limited by the availability of the experimentally determined complexed structures of effectors and host receptor binding partners, we demonstrated the potential of molecular docking simulations to predict the likely interactions between effectors and their respective host receptor binding partners. This computational approach may accelerate the process of the discovery of putative interacting plant partners of effector proteins and contribute to effector-assisted marker discovery, thereby supporting the breeding of disease-resistant crops.

ComplexQA: a deep graph learning approach for protein complex structure assessment

MOTIVATION

In recent years, the end-to-end deep learning method for single-chain protein structure prediction has achieved high accuracy. For example, the state-of-the-art method AlphaFold, developed by Google, has largely increased the accuracy of protein structure predictions to near experimental accuracy in some of the cases. At the same time, there are few methods that can evaluate the quality of protein complexes at the residue level. In particular, evaluating the quality of residues at the interface of protein complexes can lead to a wide range of applications, such as protein function analysis and drug design. In this paper, we introduce a new deep graph neural network-based method ComplexQA, to evaluate the local quality of interfaces for protein complexes by utilizing the residue-level structural information in 3D space and the sequence-level constraints.

RESULTS

We benchmark our method to other state-of-the-art quality assessment approaches on the HAF2 and DBM55-AF2 datasets (high-quality structural models predicted by AlphaFold-Multimer), and the BM5 docking dataset. The experimental results show that our proposed method achieves better or similar performance compared with other state-of-the-art methods, especially on difficult targets which only contain a few acceptable models. Our method is able to suggest a score for each interfac e residue, which demonstrates a powerful assessment tool for the ever-increasing number of protein complexes.

AVAILABILITY

https://github.com/Cao-Labs/ComplexQA.git. Contact: caora@plu.edu.

Discriminating physiological from non-physiological interfaces in structures of protein complexes: A community-wide study

Reliably scoring and ranking candidate models of protein complexes and assigning their oligomeric state from the structure of the crystal lattice represent outstanding challenges. A community-wide effort was launched to tackle these challenges. The latest resources on protein complexes and interfaces were exploited to derive a benchmark dataset consisting of 1677 homodimer protein crystal structures, including a balanced mix of physiological and non-physiological complexes. The non-physiological complexes in the benchmark were selected to bury a similar or larger interface area than their physiological counterparts, making it more difficult for scoring functions to differentiate between them. Next, 252 functions for scoring protein-protein interfaces previously developed by 13 groups were collected and evaluated for their ability to discriminate between physiological and non-physiological complexes. A simple consensus score generated using the best performing score of each of the 13 groups, and a cross-validated Random Forest (RF) classifier were created. Both approaches showed excellent performance, with an area under the Receiver Operating Characteristic (ROC) curve of 0.93 and 0.94, respectively, outperforming individual scores developed by different groups. Additionally, AlphaFold2 engines recalled the physiological dimers with significantly higher accuracy than the non-physiological set, lending support to the reliability of our benchmark dataset annotations. Optimizing the combined power of interface scoring functions and evaluating it on challenging benchmark datasets appears to be a promising strategy.

Adaptive evolution in virulence effectors of the rice blast fungus Pyricularia oryzae

Plant pathogens secrete proteins called effectors that target host cellular processes to promote disease. Recently, structural genomics has identified several families of fungal effectors that share a similar three-dimensional structure despite remarkably variable amino-acid sequences and surface properties. To explore the selective forces that underlie the sequence variability of structurally-analogous effectors, we focused on MAX effectors, a structural family of effectors that are major determinants of virulence in the rice blast fungus Pyricularia oryzae. Using structure-informed gene annotation, we identified 58 to 78 MAX effector genes per genome in a set of 120 isolates representing seven host-associated lineages. The expression of MAX effector genes was primarily restricted to the early biotrophic phase of infection and strongly influenced by the host plant. Pangenome analyses of MAX effectors demonstrated extensive presence/absence polymorphism and identified gene loss events possibly involved in host range adaptation. However, gene knock-in experiments did not reveal a strong effect on virulence phenotypes suggesting that other evolutionary mechanisms are the main drivers of MAX effector losses. MAX effectors displayed high levels of standing variation and high rates of non-synonymous substitutions, pointing to widespread positive selection shaping the molecular diversity of MAX effectors. The combination of these analyses with structural data revealed that positive selection acts mostly on residues located in particular structural elements and at specific positions. By providing a comprehensive catalog of amino acid polymorphism, and by identifying the structural determinants of the sequence diversity, our work will inform future studies aimed at elucidating the function and mode of action of MAX effectors.

PointDE: Protein Docking Evaluation Using 3D Point Cloud Neural Network

Protein-protein interactions (PPIs) play essential roles in many vital movements and the determination of protein complex structure is helpful to discover the mechanism of PPI. Protein-protein docking is being developed to model the structure of the protein. However, there is still a challenge to selecting the near-native decoys generated by protein-protein docking. Here, we propose a docking evaluation method using 3D point cloud neural network named PointDE. PointDE transforms protein structure to the point cloud. Using the state-of-the-art point cloud network architecture and a novel grouping mechanism, PointDE can capture the geometries of the point cloud and learn the interaction information from the protein interface. On public datasets, PointDE surpasses the state-of-the-art method using deep learning. To further explore the ability of our method in different types of protein structures, we developed a new dataset generated by high-quality antibody-antigen complexes. The result in this antibody-antigen dataset shows the strong performance of PointDE, which will be helpful for the understanding of PPI mechanisms.

Modeling protein-nucleic acid complexes with extremely large conformational changes using Flex-LZerD

Proteins and nucleic acids are key components in many processes in living cells, and interactions between proteins and nucleic acids are often crucial pathway components. In many cases, large flexibility of proteins as they interact with nucleic acids is key to their function. To understand the mechanisms of these processes, it is necessary to consider the 3D atomic structures of such protein-nucleic acid complexes. When such structures are not yet experimentally determined, protein docking can be used to computationally generate useful structure models. However, such docking has long had the limitation that the consideration of flexibility is usually limited to small movements or to small structures. We previously developed a method of flexible protein docking which could model ordered proteins which undergo large-scale conformational changes, which we also showed was compatible with nucleic acids. Here, we elaborate on the ability of that pipeline, Flex-LZerD, to model specifically interactions between proteins and nucleic acids, and demonstrate that Flex-LZerD can model more interactions and types of conformational change than previously shown.

Automated simulation-based membrane protein refinement into cryo-EM data

The resolution revolution has increasingly enabled single-particle cryogenic electron microscopy (cryo-EM) reconstructions of previously inaccessible systems, including membrane proteins-a category that constitutes a disproportionate share of drug targets. We present a protocol for using density-guided molecular dynamics simulations to automatically refine atomistic models into membrane protein cryo-EM maps. Using adaptive force density-guided simulations as implemented in the GROMACS molecular dynamics package, we show how automated model refinement of a membrane protein is achieved without the need to manually tune the fitting force ad hoc. We also present selection criteria to choose the best-fit model that balances stereochemistry and goodness of fit. The proposed protocol was used to refine models into a new cryo-EM density of the membrane protein maltoporin, either in a lipid bilayer or detergent micelle, and we found that results do not substantially differ from fitting in solution. Fitted structures satisfied classical model-quality metrics and improved the quality and the model-to-map correlation of the x-ray starting structure. Additionally, the density-guided fitting in combination with generalized orientation-dependent all-atom potential was used to correct the pixel-size estimation of the experimental cryo-EM density map. This work demonstrates the applicability of a straightforward automated approach to fitting membrane protein cryo-EM densities. Such computational approaches promise to facilitate rapid refinement of proteins under different conditions or with various ligands present, including targets in the highly relevant superfamily of membrane proteins.

A gated graph transformer for protein complex structure quality assessment and its performance in CASP15

MOTIVATION

Proteins interact to form complexes to carry out essential biological functions. Computational methods such as AlphaFold-multimer have been developed to predict the quaternary structures of protein complexes. An important yet largely unsolved challenge in protein complex structure prediction is to accurately estimate the quality of predicted protein complex structures without any knowledge of the corresponding native structures. Such estimations can then be used to select high-quality predicted complex structures to facilitate biomedical research such as protein function analysis and drug discovery.

RESULTS

In this work, we introduce a new gated neighborhood-modulating graph transformer to predict the quality of 3D protein complex structures. It incorporates node and edge gates within a graph transformer framework to control information flow during graph message passing. We trained, evaluated and tested the method (called DProQA) on newly-curated protein complex datasets before the 15th Critical Assessment of Techniques for Protein Structure Prediction (CASP15) and then blindly tested it in the 2022 CASP15 experiment. The method was ranked 3rd among the single-model quality assessment methods in CASP15 in terms of the ranking loss of TM-score on 36 complex targets. The rigorous internal and external experiments demonstrate that DProQA is effective in ranking protein complex structures.

AVAILABILITY AND IMPLEMENTATION

The source code, data, and pre-trained models are available at https://github.com/jianlin-cheng/DProQA.

PIQLE: protein-protein interface quality estimation by deep graph learning of multimeric interaction geometries

Motivation

Accurate modeling of protein-protein interaction interface is essential for high-quality protein complex structure prediction. Existing approaches for estimating the quality of a predicted protein complex structural model utilize only the physicochemical properties or energetic contributions of the interacting atoms, ignoring evolutionarily information or inter-atomic multimeric geometries, including interaction distance and orientations.

Results

Here, we present PIQLE, a deep graph learning method for protein-protein interface quality estimation. PIQLE leverages multimeric interaction geometries and evolutionarily information along with sequence- and structure-derived features to estimate the quality of individual interactions between the interfacial residues using a multi-head graph attention network and then probabilistically combines the estimated quality for scoring the overall interface. Experimental results show that PIQLE consistently outperforms existing state-of-the-art methods including DProQA, TRScore, GNN-DOVE and DOVE on multiple independent test datasets across a wide range of evaluation metrics. Our ablation study and comparison with the self-assessment module of AlphaFold-Multimer repurposed for protein complex scoring reveal that the performance gains are connected to the effectiveness of the multi-head graph attention network in leveraging multimeric interaction geometries and evolutionary information along with other sequence- and structure-derived features adopted in PIQLE.

Availability and implementation

An open-source software implementation of PIQLE is freely available at https://github.com/Bhattacharya-Lab/PIQLE.

Supplementary information

Supplementary data are available at Bioinformatics Advances online.

Critical Assessment of Methods for Predicting the 3D Structure of Proteins and Protein Complexes

Advances in a scientific discipline are often measured by small, incremental steps. In this review, we report on two intertwined disciplines in the protein structure prediction field, modeling of single chains and modeling of complexes, that have over decades emulated this pattern, as monitored by the community-wide blind prediction experiments CASP and CAPRI. However, over the past few years, dramatic advances were observed for the accurate prediction of single protein chains, driven by a surge of deep learning methodologies entering the prediction field. We review the mainscientific developments that enabled these recent breakthroughs and feature the important role of blind prediction experiments in building up and nurturing the structure prediction field. We discuss how the new wave of artificial intelligence-based methods is impacting the fields of computational and experimental structural biology and highlight areas in which deep learning methods are likely to lead to future developments, provided that major challenges are overcome.

PIQLE: protein-protein interface quality estimation by deep graph learning of multimeric interaction geometries

Accurate modeling of protein-protein interaction interface is essential for high-quality protein complex structure prediction. Existing approaches for estimating the quality of a predicted protein complex structural model utilize only the physicochemical properties or energetic contributions of the interacting atoms, ignoring evolutionarily information or inter-atomic multimeric geometries, including interaction distance and orientations. Here we present PIQLE, a deep graph learning method for protein-protein interface quality estimation. PIQLE leverages multimeric interaction geometries and evolutionarily information along with sequence- and structure-derived features to estimate the quality of the individual interactions between the interfacial residues using a multihead graph attention network and then probabilistically combines the estimated quality of the interfacial residues for scoring the overall interface. Experimental results show that PIQLE consistently outperforms existing state-of-the-art methods on multiple independent test datasets across a wide range of evaluation metrics. Our ablation study reveals that the performance gains are connected to the effectiveness of the multihead graph attention network in leveraging multimeric interaction geometries and evolutionary information along with other sequence- and structure-derived features adopted in PIQLE. An open-source software implementation of PIQLE, licensed under the GNU General Public License v3, is freely available at https://github.com/Bhattacharya-Lab/PIQLE .

G-RANK: an equivariant graph neural network for the scoring of protein-protein docking models

Motivation

Protein complex structure prediction is important for many applications in bioengineering. A widely used method for predicting the structure of protein complexes is computational docking. Although many tools for scoring protein-protein docking models have been developed, it is still a challenge to accurately identify near-native models for unknown protein complexes. A recently proposed model called the geometric vector perceptron-graph neural network (GVP-GNN), a subtype of equivariant graph neural networks, has demonstrated success in various 3D molecular structure modeling tasks.

Results

Herein, we present G-RANK, a GVP-GNN-based method for the scoring of protein-protein docking models. When evaluated on two different test datasets, G-RANK achieved a performance competitive with or better than the state-of-the-art scoring functions. We expect G-RANK to be a useful tool for various applications in biological engineering.

Availability and implementation

Source code is available at https://github.com/ha01994/grank.

Contact

kds@kaist.ac.kr.

Supplementary information

Supplementary data are available at Bioinformatics Advances online.

Improving classification of correct and incorrect protein-protein docking models by augmenting the training set

Motivation

Protein-protein interactions drive many relevant biological events, such as infection, replication and recognition. To control or engineer such events, we need to access the molecular details of the interaction provided by experimental 3D structures. However, such experiments take time and are expensive; moreover, the current technology cannot keep up with the high discovery rate of new interactions. Computational modeling, like protein-protein docking, can help to fill this gap by generating docking poses. Protein-protein docking generally consists of two parts, sampling and scoring. The sampling is an exhaustive search of the tridimensional space. The caveat of the sampling is that it generates a large number of incorrect poses, producing a highly unbalanced dataset. This limits the utility of the data to train machine learning classifiers.

Results

Using weak supervision, we developed a data augmentation method that we named hAIkal. Using hAIkal, we increased the labeled training data to train several algorithms. We trained and obtained different classifiers; the best classifier has 81% accuracy and 0.51 Matthews' correlation coefficient on the test set, surpassing the state-of-the-art scoring functions.

Availability and implementation

Docking models from Benchmark 5 are available at https://doi.org/10.5281/zenodo.4012018. Processed tabular data are available at https://repository.kaust.edu.sa/handle/10754/666961. Google colab is available at https://colab.research.google.com/drive/1vbVrJcQSf6\_C3jOAmZzgQbTpuJ5zC1RP?usp=sharing.

Supplementary information

Supplementary data are available at Bioinformatics Advances online.

De novo protein fold design through sequence-independent fragment assembly simulations

De novo protein design generally consists of two steps, including structure and sequence design. Many protein design studies have focused on sequence design with scaffolds adapted from native structures in the PDB, which renders novel areas of protein structure and function space unexplored. We developed FoldDesign to create novel protein folds from specific secondary structure (SS) assignments through sequence-independent replica-exchange Monte Carlo (REMC) simulations. The method was tested on 354 non-redundant topologies, where FoldDesign consistently created stable structural folds, while recapitulating on average 87.7% of the SS elements. Meanwhile, the FoldDesign scaffolds had well-formed structures with buried residues and solvent-exposed areas closely matching their native counterparts. Despite the high fidelity to the input SS restraints and local structural characteristics of native proteins, a large portion of the designed scaffolds possessed global folds completely different from natural proteins in the PDB, highlighting the ability of FoldDesign to explore novel areas of protein fold space. Detailed data analyses revealed that the major contributions to the successful structure design lay in the optimal energy force field, which contains a balanced set of SS packing terms, and REMC simulations, which were coupled with multiple auxiliary movements to efficiently search the conformational space. Additionally, the ability to recognize and assemble uncommon super-SS geometries, rather than the unique arrangement of common SS motifs, was the key to generating novel folds. These results demonstrate a strong potential to explore both structural and functional spaces through computational design simulations that natural proteins have not reached through evolution.

Pairwise and Multi-chain Protein Docking Enhanced Using LZerD Web Server

Interactions of proteins with other macromolecules have important structural and functional roles in the basic processes of living cells. To understand and elucidate the mechanisms of interactions, it is important to know the 3D structures of the complexes. Proteomes contain numerous protein-protein complexes, for which experimentally determined structures often do not exist. Computational techniques can be a practical alternative to obtain useful complex structure models. Here, we present a web server that provides access to the LZerD and Multi-LZerD protein docking tools, which can perform both pairwise and multi-chain docking. The web server is user-friendly, with options to visualize the distribution and structures of binding poses of top-scoring models. The LZerD web server is available at https://lzerd.kiharalab.org . This chapter dictates the algorithm and step-by-step procedure to model the monomeric structures with AttentiveDist, and also provides the detail of pairwise LZerD docking, and multi-LZerD. This also provided case studies for each of the three modules.

Domain-Based Protein Docking with Extremely Large Conformational Changes

Proteins are key components in many processes in living cells, and physical interactions with other proteins and nucleic acids often form key parts of their functions. In many cases, large flexibility of proteins as they interact is key to their function. To understand the mechanisms of these processes, it is necessary to consider the 3D structures of such protein complexes. When such structures are not yet experimentally determined, protein docking has long been present to computationally generate useful structure models. However, protein docking has long had the limitation that the consideration of flexibility is usually limited to very small movements or very small structures. Methods have been developed which handle minor flexibility via normal mode or other structure sampling, but new methods are required to model ordered proteins which undergo large-scale conformational changes to elucidate their function at the molecular level. Here, we present Flex-LZerD, a framework for docking such complexes. Via partial assembly multidomain docking and an iterative normal mode analysis admitting curvilinear motions, we demonstrate the ability to model the assembly of a variety of protein-protein and protein-nucleic acid complexes.

RL-MLZerD: Multimeric protein docking using reinforcement learning

Numerous biological processes in a cell are carried out by protein complexes. To understand the molecular mechanisms of such processes, it is crucial to know the quaternary structures of the complexes. Although the structures of protein complexes have been determined by biophysical experiments at a rapid pace, there are still many important complex structures that are yet to be determined. To supplement experimental structure determination of complexes, many computational protein docking methods have been developed; however, most of these docking methods are designed only for docking with two chains. Here, we introduce a novel method, RL-MLZerD, which builds multiple protein complexes using reinforcement learning (RL). In RL-MLZerD a multi-chain assembly process is considered as a series of episodes of selecting and integrating pre-computed pairwise docking models in a RL framework. RL is effective in correctly selecting plausible pairwise models that fit well with other subunits in a complex. When tested on a benchmark dataset of protein complexes with three to five chains, RL-MLZerD showed better modeling performance than other existing multiple docking methods under different evaluation criteria, except against AlphaFold-Multimer in unbound docking. Also, it emerged that the docking order of multi-chain complexes can be naturally predicted by examining preferred paths of episodes in the RL computation.

Estimation of model accuracy by a unique set of features and tree-based regressor

Computationally generated models of protein structures bridge the gap between the practically negligible price tag of sequencing and the high cost of experimental structure determination. By providing a low-cost (and often free) partial alternative to experimentally determined structures, these models help biologists design and interpret their experiments. Obviously, the more accurate the models the more useful they are. However, methods for protein structure prediction generate many structural models of various qualities, necessitating means for the estimation of their accuracy. In this work we present MESHI_consensus, a new method for the estimation of model accuracy. The method uses a tree-based regressor and a set of structural, target-based, and consensus-based features. The new method achieved high performance in the EMA (Estimation of Model Accuracy) track of the recent CASP14 community-wide experiment (https://predictioncenter.org/casp14/index.cgi). The tertiary structure prediction track of that experiment revealed an unprecedented leap in prediction performance by a single prediction group/method, namely AlphaFold2. This achievement would inevitably have a profound impact on the field of protein structure prediction, including the accuracy estimation sub-task. We conclude this manuscript with some speculations regarding the future role of accuracy estimation in a new era of accurate protein structure prediction.

Multi-level analysis of intrinsically disordered protein docking methods

Intrinsically Disordered Proteins (IDPs) are a class of proteins in which at least some region of the protein does not possess any stable structure in solution in the physiological condition but may adopt an ordered structure upon binding to a globular receptor. These IDP-receptor complexes are thus subject to protein complex modeling in which computational techniques are applied to accurately reproduce the IDP ligand-receptor interactions. This often exists in the form of protein docking, in which the 3D structures of both the subunits are known, but the position of the ligand relative to the receptor is not. Here, we evaluate the performance of three IDP-receptor modeling tools with metrics that characterize the IDP-receptor interface at various resolutions. We show that all three methods are able to properly identify the general binding site, as identified by lower resolution metrics, but begin to struggle with higher resolution metrics that capture biophysical interactions.

Protein Model Quality Estimation Using Molecular Dynamics Simulation

The estimation of protein model quality remains a challenging task and is important for protein structural model utilization. In the last decade, existing methods that rely on machine learning to deep learning have been developed and shown progressive improvement. Despite utilizing more sophisticated techniques and introducing new features, none of these methods employ explicit protein structure stability information. Hypothetically, protein model quality might be indicated by its structural stability in an in silico system disclosed by the structural difference from its initial structure. One of the possible methods to exploit such information is by implementing molecular dynamics simulations that have shown successful applications in many research fields. We present a novel approach by introducing explicit protein structure stability information using molecular dynamics simulation. Despite using only simple features, small data with no training process required, and a short molecular dynamics simulation time, our method shows comparable performance to the state-of-the-art deep learning-based method.

Structure Prediction, Evaluation, and Validation of GPR18 Lipid Receptor Using Free Programs

The GPR18 receptor, often referred to as the N-arachidonylglycine receptor, although assigned (along with GPR55 and GPR119) to the new class A GPCR subfamily-lipid receptors, officially still has the status of a class A GPCR orphan. While its signaling pathways and biological significance have not yet been fully elucidated, increasing evidence points to the therapeutic potential of GPR18 in relation to immune, neurodegenerative, and cancer processes to name a few. Therefore, it is necessary to understand the interactions of potential ligands with the receptor and the influence of particular structural elements on their activity. Thus, given the lack of an experimentally solved structure, the goal of the present study was to obtain a homology model of the GPR18 receptor in the inactive state, meeting all requirements in terms of protein structure quality and recognition of active ligands. To increase the reliability and precision of the predictions, different contemporary protein structure prediction methods and software were used and compared herein. To test the usability of the resulting models, we optimized and compared the selected structures followed by the assessment of the ability to recognize known, active ligands. The stability of the predicted poses was then evaluated by means of molecular dynamics simulations. On the other hand, most of the best-ranking contemporary CADD software/platforms for its full usability require rather expensive licenses. To overcome this down-to-earth obstacle, the overarching goal of these studies was to test whether it is possible to perform the thorough CADD experiments with high scientific confidence while using only license-free/academic software and online platforms. The obtained results indicate that a wide range of freely available software and/or academic licenses allow us to carry out meaningful molecular modelling/docking studies.

Multiple-Allele MHC Class II Epitope Engineering by a Molecular Dynamics-Based Evolution Protocol

Epitopes that bind simultaneously to all human alleles of Major Histocompatibility Complex class II (MHC II) are considered one of the key factors for the development of improved vaccines and cancer immunotherapies. To engineer MHC II multiple-allele binders, we developed a protocol called PanMHC-PARCE, based on the unsupervised optimization of the epitope sequence by single-point mutations, parallel explicit-solvent molecular dynamics simulations and scoring of the MHC II-epitope complexes. The key idea is accepting mutations that not only improve the affinity but also reduce the affinity gap between the alleles. We applied this methodology to enhance a Plasmodium vivax epitope for multiple-allele binding. In vitro rate-binding assays showed that four engineered peptides were able to bind with improved affinity toward multiple human MHC II alleles. Moreover, we demonstrated that mice immunized with the peptides exhibited interferon-gamma cellular immune response. Overall, the method enables the engineering of peptides with improved binding properties that can be used for the generation of new immunotherapies.

Improved Protein Decoy Selection via Non-Negative Matrix Factorization

A central challenge in protein modeling research and protein structure prediction in particular is known as decoy selection. The problem refers to selecting biologically-active/native tertiary structures among a multitude of physically-realistic structures generated by template-free protein structure prediction methods. Research on decoy selection is active. Clustering-based methods are popular, but they fail to identify good/near-native decoys on datasets where near-native decoys are severely under-sampled by a protein structure prediction method. Reasonable progress is reported by methods that additionally take into account the internal energy of a structure and employ it to identify basins in the energy landscape organizing the multitude of decoys. These methods, however, incur significant time costs for extracting basins from the landscape. In this paper, we propose a novel decoy selection method based on non-negative matrix factorization. We demonstrate that our method outperforms energy landscape-based methods. In particular, the proposed method addresses both the time cost issue and the challenge of identifying good decoys in a sparse dataset, successfully recognizing near-native decoys for both easy and hard protein targets.

Structure‐conditioned amino‐acid couplings: how contact geometry affects pairwise sequence preferences

Relating a protein's sequence to its conformation is a central challenge for both structure prediction and sequence design. Statistical contact potentials, as well as their more descriptive versions that account for side‐chain orientation and other geometric descriptors, have served as simplistic but useful means of representing second‐order contributions in sequence–structure relationships. Here we ask what happens when a pairwise potential is conditioned on the fully defined geometry of interacting backbones fragments. We show that the resulting structure‐conditioned coupling energies more accurately reflect pair preferences as a function of structural contexts. These structure‐conditioned energies more reliably encode native sequence information and more highly correlate with experimentally determined coupling energies. Clustering a database of interaction motifs by structure results in ensembles of similar energies and clustering them by energy results in ensembles of similar structures. By comparing many pairs of interaction motifs and showing that structural similarity and energetic similarity go hand‐in‐hand, we provide a tangible link between modular sequence and structure elements. This link is applicable to structural modeling, and we show that scoring CASP models with structured‐conditioned energies results in substantially higher correlation with structural quality than scoring the same models with a contact potential. We conclude that structure‐conditioned coupling energies are a good way to model the impact of interaction geometry on second‐order sequence preferences.

Computational Evolution Protocol for Peptide Design

Computational peptide design is useful for therapeutics, diagnostics, and vaccine development. To select the most promising peptide candidates, the key is describing accurately the peptide-target interactions at the molecular level. We here review a computational peptide design protocol whose key feature is the use of all-atom explicit solvent molecular dynamics for describing the different peptide-target complexes explored during the optimization. We describe the milestones behind the development of this protocol, which is now implemented in an open-source code called PARCE. We provide a basic tutorial to run the code for an antibody fragment design example. Finally, we describe three additional applications of the method to design peptides for different targets, illustrating the broad scope of the proposed approach.

Benchmarking of structure refinement methods for protein complex models

Protein structure docking is the process in which the quaternary structure of a protein complex is predicted from individual tertiary structures of the protein subunits. Protein docking is typically performed in two main steps. The subunits are first docked while keeping them rigid to form the complex, which is then followed by structure refinement. Structure refinement is crucial for a practical use of computational protein docking models, as it is aimed for correcting conformations of interacting residues and atoms at the interface. Here, we benchmarked the performance of eight existing protein structure refinement methods in refinement of protein complex models. We show that the fraction of native contacts between subunits is by far the most straightforward metric to improve. However, backbone dependent metrics, based on the Root Mean Square Deviation proved more difficult to improve via refinement.

A random forest classifier for protein-protein docking models

Herein, we present the results of a machine learning approach we developed to single out correct 3D docking models of protein-protein complexes obtained by popular docking software. To this aim, we generated 3 × 10 4 docking models for each of the 230 complexes in the protein-protein benchmark, version 5, using three different docking programs (HADDOCK, FTDock and ZDOCK), for a cumulative set of ≈ 7 × 10 6 docking models. Three different machine learning approaches (Random Forest, Supported Vector Machine and Perceptron) were used to train classifiers with 158 different scoring functions (features). The Random Forest algorithm outperformed the other two algorithms and was selected for further optimization. Using a features selection algorithm, and optimizing the random forest hyperparameters, allowed us to train and validate a random forest classifier, named COnservation Driven Expert System (CoDES). Testing of CoDES on independent datasets, as well as results of its comparative performance with machine learning methods recently developed in the field for the scoring of docking decoys, confirm its state-of-the-art ability to discriminate correct from incorrect decoys both in terms of global parameters and in terms of decoys ranked at the top positions.

Supplementary information

Supplementary data are available at Bioinformatics Advances online.

Software and data availability statement

The docking models are available at https://doi.org/10.5281/zenodo.4012018. The programs underlying this article will be shared on request to the corresponding authors.

Virtual Screening with Gnina 1.0

Virtual screening-predicting which compounds within a specified compound library bind to a target molecule, typically a protein-is a fundamental task in the field of drug discovery. Doing virtual screening well provides tangible practical benefits, including reduced drug development costs, faster time to therapeutic viability, and fewer unforeseen side effects. As with most applied computational tasks, the algorithms currently used to perform virtual screening feature inherent tradeoffs between speed and accuracy. Furthermore, even theoretically rigorous, computationally intensive methods may fail to account for important effects relevant to whether a given compound will ultimately be usable as a drug. Here we investigate the virtual screening performance of the recently released Gnina molecular docking software, which uses deep convolutional networks to score protein-ligand structures. We find, on average, that Gnina outperforms conventional empirical scoring. The default scoring in Gnina outperforms the empirical AutoDock Vina scoring function on 89 of the 117 targets of the DUD-E and LIT-PCBA virtual screening benchmarks with a median 1% early enrichment factor that is more than twice that of Vina. However, we also find that issues of bias linger in these sets, even when not used directly to train models, and this bias obfuscates to what extent machine learning models are achieving their performance through a sophisticated interpretation of molecular interactions versus fitting to non-informative simplistic property distributions.

Prediction of protein assemblies, the next frontier: The CASP14-CAPRI experiment

We present the results for CAPRI Round 50, the fourth joint CASP-CAPRI protein assembly prediction challenge. The Round comprised a total of twelve targets, including six dimers, three trimers, and three higher-order oligomers. Four of these were easy targets, for which good structural templates were available either for the full assembly, or for the main interfaces (of the higher-order oligomers). Eight were difficult targets for which only distantly related templates were found for the individual subunits. Twenty-five CAPRI groups including eight automatic servers submitted ~1250 models per target. Twenty groups including six servers participated in the CAPRI scoring challenge submitted ~190 models per target. The accuracy of the predicted models was evaluated using the classical CAPRI criteria. The prediction performance was measured by a weighted scoring scheme that takes into account the number of models of acceptable quality or higher submitted by each group as part of their five top-ranking models. Compared to the previous CASP-CAPRI challenge, top performing groups submitted such models for a larger fraction (70-75%) of the targets in this Round, but fewer of these models were of high accuracy. Scorer groups achieved stronger performance with more groups submitting correct models for 70-80% of the targets or achieving high accuracy predictions. Servers performed less well in general, except for the MDOCKPP and LZERD servers, who performed on par with human groups. In addition to these results, major advances in methodology are discussed, providing an informative overview of where the prediction of protein assemblies currently stands.

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.

Predicting protein interaction network perturbation by alternative splicing with semi-supervised learning

Alternative splicing introduces an additional layer of protein diversity and complexity in regulating cellular functions that can be specific to the tissue and cell type, physiological state of a cell, or disease phenotype. Recent high-throughput experimental studies have illuminated the functional role of splicing events through rewiring protein-protein interactions; however, the extent to which the macromolecular interactions are affected by alternative splicing has yet to be fully understood. In silico methods provide a fast and cheap alternative to interrogating functional characteristics of thousands of alternatively spliced isoforms. Here, we develop an accurate feature-based machine learning approach that predicts whether a protein-protein interaction carried out by a reference isoform is perturbed by an alternatively spliced isoform. Our method, called the alternatively spliced interactions prediction (ALT-IN) tool, is compared with the state-of-the-art PPI prediction tools and shows superior performance, achieving 0.92 in precision and recall values.

LZerD Protein-Protein Docking Webserver Enhanced With de novo Structure Prediction

Protein-protein docking is a useful tool for modeling the structures of protein complexes that have yet to be experimentally determined. Understanding the structures of protein complexes is a key component for formulating hypotheses in biophysics regarding the functional mechanisms of complexes. Protein-protein docking is an established technique for cases where the structures of the subunits have been determined. While the number of known structures deposited in the Protein Data Bank is increasing, there are still many cases where the structures of individual proteins that users want to dock are not determined yet. Here, we have integrated the AttentiveDist method for protein structure prediction into our LZerD webserver for protein-protein docking, which enables users to simply submit protein sequences and obtain full-complex atomic models, without having to supply any structure themselves. We have further extended the LZerD docking interface with a symmetrical homodimer mode. The LZerD server is available at https://lzerd.kiharalab.org/.

Empirical Potential Energy Function Toward ab Initio Folding G Protein-Coupled Receptors

Approximately 40-50 percent of all drugs targets are G protein-coupled receptors (GPCRs). Three-dimensional structure of GPCRs is important to probe their biophysical and biochemical functions and their pharmaceutical applications. Lacking reliable and high quality free function is one of the ugent problems of computational predicting the three-dimensional structure in this community. We proposed a GPCR-specified energy function composed of four novel empirical potential energy terms: a two-dimensional contact energy force field, knowledge-based helix pair connection distance energy term, knowledge-based helix pair angle restraint energy term and a disulfide bond energy term. To validate the energy function, we employed an ab initio GPCR three-dimensional structure predictor to test if the energy function improved the accuracy of prediction. We evaluated 28 solved GPCRs and found that 21(75 percent) targets were correctly folded (TM-score>0.5). Also, the average TM-score using the energy function was 0.54, which was improved 134 percent than the TM-score 0.23 for MODELLER energy function and 170 percent than the TM-score 0.20 for Rosetta membrane energy function. The results confirmed that our empirical potential energy function toward ab initio folding is competitive to state-of-the-art solutions for structural prediction of GPCRs.

Assessment of protein model structure accuracy estimation in CASP14: Old and new challenges

In CASP, blind testing of model accuracy estimation methods has been conducted on models submitted by tertiary structure prediction servers. In CASP14, model accuracy estimation results were evaluated in terms of both global and local structure accuracy, as in the previous CASPs. Unlike the previous CASPs that did not show pronounced improvements in performance, the best single-model method (from the Baker group) showed an improved performance in CASP14, particularly in evaluating global structure accuracy when compared to both the best single-model methods in previous CASPs and the best multi-model methods in the current CASP. Although the CASP14 experiment on model accuracy estimation did not deal with the structures generated by AlphaFold2, new challenges that have arisen due to the success of AlphaFold2 are discussed.

Predicting mutant outcome by combining deep mutational scanning and machine learning

Deep mutational scanning provides unprecedented wealth of quantitative data regarding the functional outcome of mutations in proteins. A single experiment may measure properties (eg, structural stability) of numerous protein variants. Leveraging the experimental data to gain insights about unexplored regions of the mutational landscape is a major computational challenge. Such insights may facilitate further experimental work and accelerate the development of novel protein variants with beneficial therapeutic or industrially relevant properties. Here we present a novel, machine learning approach for the prediction of functional mutation outcome in the context of deep mutational screens. Using sequence (one-hot) features of variants with known properties, as well as structural features derived from models thereof, we train predictive statistical models to estimate the unknown properties of other variants. The utility of the new computational scheme is demonstrated using five sets of mutational scanning data, denoted "targets": (a) protease specificity of APPI (amyloid precursor protein inhibitor) variants; (b-d) three stability related properties of IGBPG (immunoglobulin G-binding β1 domain of streptococcal protein G) variants; and (e) fluorescence of GFP (green fluorescent protein) variants. Performance is measured by the overall correlation of the predicted and observed properties, and enrichment-the ability to predict the most potent variants and presumably guide further experiments. Despite the diversity of the targets the statistical models can generalize variant examples thereof and predict the properties of test variants with both single and multiple mutations.

LZerD webserver for pairwise and multiple protein-protein docking

Protein complexes are involved in many important processes in living cells. To understand the mechanisms of these processes, it is necessary to solve the 3D structures of the protein complexes. When protein complex structures have not yet been determined by experiment, protein-protein docking tools can be used to computationally model the structures of these complexes. Here, we present a webserver which provides access to LZerD and Multi-LZerD protein docking tools. The protocol provided by the server have performed consistently among the top in the CAPRI blind evaluation. LZerD docks pairs of structures, while Multi-LZerD can dock three or more structures simultaneously. LZerD uses a soft protein surface representation with 3D Zernike descriptors and explores the binding pose space using geometric hashing. Multi-LZerD performs multi-chain docking by combining pairwise solutions by LZerD. Both methods output full-atom docked models of the input proteins. Users can also input distance constraints between interacting or non-interacting residues as well as residues that locate at the interface or far from the interface. The webserver is equipped with a user-friendly panel that visualizes the distribution and structures of binding poses of top scoring models. The LZerD webserver is available at https://lzerd.kiharalab.org.

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/.

DeepRefiner: high-accuracy protein structure refinement by deep network calibration

The DeepRefiner webserver, freely available at http://watson.cse.eng.auburn.edu/DeepRefiner/, is an interactive and fully configurable online system for high-accuracy protein structure refinement. Fuelled by deep learning, DeepRefiner offers the ability to leverage cutting-edge deep neural network architectures which can be calibrated for on-demand selection of adventurous or conservative refinement modes targeted at degree or consistency of refinement. The method has been extensively tested in the Critical Assessment of Techniques for Protein Structure Prediction (CASP) experiments under the group name 'Bhattacharya-Server' and was officially ranked as the No. 2 refinement server in CASP13 (second only to 'Seok-server' and outperforming all other refinement servers) and No. 2 refinement server in CASP14 (second only to 'FEIG-S' and outperforming all other refinement servers including 'Seok-server'). The DeepRefiner web interface offers a number of convenient features, including (i) fully customizable refinement job submission and validation; (ii) automated job status update, tracking, and notifications; (ii) interactive and interpretable web-based results retrieval with quantitative and visual analysis and (iv) extensive help information on job submission and results interpretation via web-based tutorial and help tooltips.

Toward the solution of the protein structure prediction problem

Since Anfinsen demonstrated that the information encoded in a protein's amino acid sequence determines its structure in 1973, solving the protein structure prediction problem has been the Holy Grail of structural biology. The goal of protein structure prediction approaches is to utilize computational modeling to determine the spatial location of every atom in a protein molecule starting from only its amino acid sequence. Depending on whether homologous structures can be found in the Protein Data Bank (PDB), structure prediction methods have been historically categorized as template-based modeling (TBM) or template-free modeling (FM) approaches. Until recently, TBM has been the most reliable approach to predicting protein structures, and in the absence of reliable templates, the modeling accuracy sharply declines. Nevertheless, the results of the most recent community-wide assessment of protein structure prediction experiment (CASP14) have demonstrated that the protein structure prediction problem can be largely solved through the use of end-to-end deep machine learning techniques, where correct folds could be built for nearly all single-domain proteins without using the PDB templates. Critically, the model quality exhibited little correlation with the quality of available template structures, as well as the number of sequence homologs detected for a given target protein. Thus, the implementation of deep-learning techniques has essentially broken through the 50-year-old modeling border between TBM and FM approaches and has made the success of high-resolution structure prediction significantly less dependent on template availability in the PDB library.

Protein Docking Model Evaluation by Graph Neural Networks

Physical interactions of proteins play key functional roles in many important cellular processes. To understand molecular mechanisms of such functions, it is crucial to determine the structure of protein complexes. To complement experimental approaches, which usually take a considerable amount of time and resources, various computational methods have been developed for predicting the structures of protein complexes. In computational modeling, one of the challenges is to identify near-native structures from a large pool of generated models. Here, we developed a deep learning-based approach named Graph Neural Network-based DOcking decoy eValuation scorE (GNN-DOVE). To evaluate a protein docking model, GNN-DOVE extracts the interface area and represents it as a graph. The chemical properties of atoms and the inter-atom distances are used as features of nodes and edges in the graph, respectively. GNN-DOVE was trained, validated, and tested on docking models in the Dockground database and further tested on a combined dataset of Dockground and ZDOCK benchmark as well as a CAPRI scoring dataset. GNN-DOVE performed better than existing methods, including DOVE, which is our previous development that uses a convolutional neural network on voxelized structure models.

Convolutional Neural Network-based Virtual Screening

Virtual screening is an important means for lead compound discovery. The scoring function is the key to selecting hit compounds. Many scoring functions are currently available; however, there are no all-purpose scoring functions because different scoring functions tend to have conflicting results. Recently, neural networks, especially convolutional neural networks, have constantly been penetrating drug design and most CNN-based virtual screening methods are superior to traditional docking methods, such as Dock and AutoDock. CNNbased virtual screening is expected to improve the previous model of overreliance on computational chemical screening. Utilizing the powerful learning ability of neural networks provides us with a new method for evaluating compounds. We review the latest progress of CNN-based virtual screening and propose prospects.