Representation Of Protein Structure Research Articles

Entropy-based distance cutoff for protein internal contact networks.

Protein structure networks (PSNs) have long been used to provide a coarse yet meaningful representation of protein structure, dynamics, and internal communication pathways. An important question is what criteria should be applied to construct the network so that to include relevant interresidue contacts while avoiding unnecessary connections. To address this issue, we systematically considered varying residue distance cutoff length and the probability threshold for contact formation to construct PSNs based on atomistic molecular dynamics in order to assess the amount of mutual information within the resulting representations. We found that the minimum in mutual information is universally achieved at the cutoff length of 5Å, irrespective of the applied contact formation probability threshold in all considered, distinct proteins. Assuming that the optimal PSNs should be characterized by the least amount of redundancy, which corresponds to the minimum in mutual information, this finding suggests an objective criterion for cutoff distance and supports the existing preference towards its customary selection around 5Å length, typically based to date on heuristic criteria.

Representations of protein structure for exploring the conformational space: A speed–accuracy trade-off

The recent breakthrough in the field of protein structure prediction shows the relevance of using knowledge-based based scoring functions in combination with a low-resolution 3D representation of protein macromolecules. The choice of not using all atoms is barely supported by any data in the literature, and is mostly motivated by empirical and practical reasons, such as the computational cost of assessing the numerous folds of the protein conformational space. Here, we present a comprehensive study, carried on a large and balanced benchmark of predicted protein structures, to see how different types of structural representations rank in either accuracy or calculation speed, and which ones offer the best compromise between these two criteria. We tested ten representations, including low-resolution, high-resolution, and coarse-grained approaches. We also investigated the generalization of the findings to other formalisms than the widely-used “potential of mean force” (PMF) method. Thus, we observed that representing protein structures by their β carbons—combined or not with Cα—provides the best speed–accuracy trade-off, when using a “total information gain” scoring function. For statistical PMFs, using MARTINI backbone and side-chains beads is the best option. Finally, we also demonstrated the necessity of training the reference state on all atom types, and of including the Cα atoms of glycine residues, in a Cβ-based representation.

Disclosing Allostery Through Protein Contact Networks.

Proteins are located in the twilight zone between chemistry and biology, where a peculiar kind of complexity starts. Proteins are the smallest 'devices' showing a sensible adaptation to their environment by the production of appropriate behavior when facing a specific stimulus. This fact qualifies (from the 'effector' side) proteins as nanomachines working as catalysts, motors, or switches. However (from the sensor side), the need to single out the 'specific stimulus' out of thermal noise qualifies proteins as information processing devices. Allostery corresponds to the modification of the configuration (in a broad sense) of the protein molecule in response to a specific stimulus in a non-strictly local way, thereby connecting the sensor and effector sides of the nanomachine. This is why the 'disclosing' of allostery phenomenon is at the very heart of protein function; in this chapter, we will demonstrate how a network-based representation of protein structure in terms of nodes (aminoacid residues) and edges (effective contacts between residues) is the natural language for getting rid of allosteric phenomena and, more in general, of protein structure/function relationships.

Simplified geometric representations of protein structures identify complementary interaction interfaces.

Protein‐protein interactions are critical to protein function, but three‐dimensional (3D) arrangements of interacting proteins have proven hard to predict, even given the identities and 3D structures of the interacting partners. Specifically, identifying the relevant pairwise interaction surfaces remains difficult, often relying on shape complementarity with molecular docking while accounting for molecular motions to optimize rigid 3D translations and rotations. However, such approaches can be computationally expensive, and faster, less accurate approximations may prove useful for large‐scale prediction and assembly of 3D structures of multi‐protein complexes. We asked if a reduced representation of protein geometry retains enough information about molecular properties to predict pairwise protein interaction interfaces that are tolerant of limited structural rearrangements. Here, we describe a reduced representation of 3D protein accessible surfaces on which molecular properties such as charge, hydrophobicity, and evolutionary rate can be easily mapped, implemented in the MorphProt package. Pairs of surfaces are compared to rapidly assess partner‐specific potential surface complementarity. On two available benchmarks of 185 overall known protein complexes, we observe predictions comparable to other structure‐based tools at correctly identifying protein interaction surfaces. Furthermore, we examined the effect of molecular motion through normal mode simulation on a benchmark receptor‐ligand pair and observed no marked loss of predictive accuracy for distortions of up to 6 Å Cα‐RMSD. Thus, a shape reduction of protein surfaces retains considerable information about surface complementarity, offers enhanced speed of comparison relative to more complex geometric representations, and exhibits tolerance to conformational changes.

High-performance transformation of protein structure representation from internal to Cartesian coordinates.

We present a highly parallel algorithm to convert internal coordinates of a polymeric molecule into Cartesian coordinates. Traditionally, converting the structures of polymers (e.g., proteins) from internal to Cartesian coordinates has been performed serially, due to an inherent linear dependency along the polymer chain. We show this dependency can be removed using a tree-based concatenation of coordinate transforms between segments, and then parallelized efficiently on graphics processing units (GPUs). The conversion algorithm is applicable to protein engineering and fitting protein structures to experimental data, and we observe an order of magnitude speedup using parallel processing on a GPU compared to serial execution on a CPU.

PRIGSA2: Improved version of protein repeat identification by graph spectral analysis

Tandemly repeated structural motifs in proteins form highly stable structural folds and provide multiple binding sites associated with diverse functional roles. The tertiary structure and function of these proteins are determined by the type and copy number of the repeating units. Each repeat type exhibits a unique pattern of intra- and inter-repeat unit interactions that is well-captured by the topological features in the network representation of protein structures. Here we present an improved version of our graph based algorithm, PRIGSA, with structure-based validation and filtering steps incorporated for accurate detection of tandem structural repeats. The algorithm integrates available knowledge on repeat families with de novo prediction to detect repeats in single monomer chains as well as in multimeric protein complexes. Three levels of performance evaluation are presented: comparison with state-of-the-art algorithms on benchmark dataset of repeat and nonrepeat proteins, accuracy in the detection of members of 13 known repeat families reported in UniProt and execution on the complete Protein Data Bank to show its ability to identify previously uncharacterized proteins. A ~3-fold increase in the coverage of the members of 13 known families and 3408 novel uncharacterized structural repeat proteins are identified on executing it on PDB. PRIGSA2 is available at http:// bioinf.iiit.ac.in/PRIGSA2/.

Deciphering interaction fingerprints from protein molecular surfaces using geometric deep learning.

Predicting interactions between proteins and other biomolecules solely based on structure remains a challenge in biology. A high-level representation of protein structure, the molecular surface, displays patterns of chemical and geometric features that fingerprint a protein's modes of interactions with other biomolecules. We hypothesize that proteins participating in similar interactions may share common fingerprints, independent of their evolutionary history. Fingerprints may be difficult to grasp by visual analysis but could be learned from large-scale datasets. We present MaSIF (molecular surface interaction fingerprinting), a conceptual framework based on a geometric deep learning method to capture fingerprints that are important for specific biomolecular interactions. We showcase MaSIF with three prediction challenges: protein pocket-ligand prediction, protein-protein interaction site prediction and ultrafast scanning of protein surfaces for prediction of protein-protein complexes. We anticipate that our conceptual framework will lead to improvements in our understanding of protein function and design.

Building maps of protein structure spaces in template-free protein structure prediction.

An important goal in template-free protein structure prediction is how to control the quality of computed tertiary structures of a target amino-acid sequence. Despite great advances in algorithmic research, given the size, dimensionality, and inherent characteristics of the protein structure space, this task remains exceptionally challenging. It is current practice to aim to generate as many structures as can be afforded so as to increase the likelihood that some of them will reside near the sought but unknown biologically-active/native structure. When operating within a given computational budget, this is impractical and uninformed by any metrics of interest. In this paper, we propose instead to equip algorithms that generate tertiary structures, also known as decoy generation algorithms, with memory of the protein structure space that they explore. Specifically, we propose an evolving, granularity-controllable map of the protein structure space that makes use of low-dimensional representations of protein structures. Evaluations on diverse target sequences that include recent hard CASP targets show that drastic reductions in storage can be made without sacrificing decoy quality. The presented results make the case that integrating a map of the protein structure space is a promising mechanism to enhance decoy generation algorithms in template-free protein structure prediction.

In silico prediction of protein flexibility with local structure approach

Flexibility is an intrinsic essential feature of protein structures, directly linked to their functions. To this day, most of the prediction methods use the crystallographic data (namely B-factors) as the only indicator of protein's inner flexibility and predicts them as rigid or flexible.PredyFlexy stands differently from other approaches as it relies on the definition of protein flexibility (i) not only taken from crystallographic data, but also (ii) from Root Mean Square Fluctuation (RMSFs) observed in Molecular Dynamics simulations. It also uses a specific representation of protein structures, named Long Structural Prototypes (LSPs). From Position-Specific Scoring Matrix, the 120 LSPs are predicted with a good accuracy and directly used to predict (i) the protein flexibility in three categories (flexible, intermediate and rigid), (ii) the normalized B-factors, (iii) the normalized RMSFs, and (iv) a confidence index. Prediction accuracy among these three classes is equivalent to the best two class prediction methods, while the normalized B-factors and normalized RMSFs have a good correlation with experimental and in silico values. Thus, PredyFlexy is a unique approach, which is of major utility for the scientific community. It support parallelization features and can be run on a local cluster using multiple cores.

Atomic Motif Recognition in (Bio)Polymers: Benchmarks From the Protein Data Bank.

Rationalizing the structure and structure–property relations for complex materials such as polymers or biomolecules relies heavily on the identification of local atomic motifs, e.g., hydrogen bonds and secondary structure patterns, that are seen as building blocks of more complex supramolecular and mesoscopic structures. Over the past few decades, several automated procedures have been developed to identify these motifs in proteins given the atomic structure. Being based on a very precise understanding of the specific interactions, these heuristic criteria formulate the question in a way that implies the answer, by defining a list of motifs based on those that are known to be naturally occurring. This makes them less likely to identify unexpected phenomena, such as the occurrence of recurrent motifs in disordered segments of proteins, and less suitable to be applied to different polymers whose structure is not driven by hydrogen bonds, or even to polypeptides when appearing in unusual, non-biological conditions. Here we discuss how unsupervised machine learning schemes can be used to recognize patterns based exclusively on the frequency with which different motifs occur, taking high-resolution structures from the Protein Data Bank as benchmarks. We first discuss the application of a density-based motif recognition scheme in combination with traditional representations of protein structure (namely, interatomic distances and backbone dihedrals). Then, we proceed one step further toward an entirely unbiased scheme by using as input a structural representation based on the atomic density and by employing supervised classification to objectively assess the role played by the representation in determining the nature of atomic-scale patterns.

Protein dynamics analysis reveals that missense mutations in cancer-related genes appear frequently on hinge-neighboring residues.

Missense mutations have various effects on protein structures, also leading to distorted protein dynamics that plausibly affects the function. We hypothesized that missense mutations in cancer-related genes selectively target hinge-neighboring residues that orchestrate collective structural dynamics. To test our hypothesis, we selected 69 cancer-related genes from the Cancer Gene Census database and their representative protein structures from the Protein Data Bank. We first identified the hinge residues in two global modes of motion by applying the Gaussian Network Model. We then showed that missense mutations are significantly enriched on hinge-neighboring residues in oncogenes and tumor suppressor genes. We observed that several oncogenes (eg, MAP2K1, PTPN11, and KRAS) and tumor suppressor genes (eg, EZH2, CDKN2C, and RHOA) strongly exhibit this phenomenon. This study highlights and rationalizes the functional importance of missense mutations on hinge-neighboring residues in cancer.

Identification of a new nucleotide binding site by structural alignment and site directed mutagenesis.

Nucleotide binding proteins are involved in many important cellular processes and form one of the largest protein families. Traditionally, the identification of nucleotide binding motif, such as the ATP binding P-loop, has relied on the comparison of protein sequences, consideration of the function of each of the proteins and the identification of signature motifs within the sequence. Sometimes, it is difficult to identify nucleotide binding proteins based on sequence alignment because of increased evolutionary distances. In such cases, structural alignments can provide a better guide for comparing specific features of sequences because the overall structures of these motifs are conserved despite low sequence identity. In the present study, on the basis of bioinformatics and structural comparison of three representative protein structures of Ham1 superfamily, YjjX, YggV, and YhdE, previously identified as nucleotide binding proteins, we have identified a novel nucleotide binding motif (T/SXXXXK/R). The importance of this signature motif in binding of nucleotides was validated using site directed mutagenesis. Mutations of conserved residues of the loop either decreased or completely abolished the nucleotide binding activity of the protein. We used the conserved motif identified in the study to search for other proteins having a similar motif. Two proteins, GTP cyclohydrolase II and dephospho-CoA pyrophosphorylase showed presence of the loop, suggesting that this nucleotide binding motif is not unique in the Ham1 superfamily, but represents a novel NTP recognition motif.

Optimizing the atom types of proteins through iterative knowledge-based potentials**Project supported by the National Natural Science Foundation of China (Grant No. 31670724), the National Key Research and Development Program of China (Grant Nos. 2016YFC1305800 and 2016YFC1305805), and the Startup Grant of Huazhong University of Science and Technology, China.

Knowledge-based scoring functions have been widely used for protein structure prediction, protein–small molecule, and protein–nucleic acid interactions, in which one critical step is to find an appropriate representation of protein structures. A key issue is to determine the minimal protein representations, which is important not only for developing of scoring functions but also for understanding the physics of protein folding. Despite significant progresses in simplifying residues into alphabets, few studies have been done to address the optimal number of atom types for proteins. Here, we have investigated the atom typing issue by classifying the 167 heavy atoms of proteins through 11 schemes with 1 to 20 atom types based on their physicochemical and functional environments. For each atom typing scheme, a statistical mechanics-based iterative method was used to extract atomic distance-dependent potentials from protein structures. The atomic distance-dependent pair potentials for different schemes were illustrated by several typical atom pairs with different physicochemical properties. The derived potentials were also evaluated on a high-resolution test set of 148 diverse proteins for native structure recognition. It was found that there was a crossover around the scheme of four atom types in terms of the success rate as a function of the number of atom types, which means that four atom types may be used when investigating the basic folding mechanism of proteins. However, it was revealed by a close examination of typical potentials that 14 atom types were needed to describe the protein interactions at atomic level. The present study will be beneficial for the development of protein related scoring functions and the understanding of folding mechanisms.

New Deep Learning Methods for Protein Loop Modeling.

Computational protein structure prediction is a long-standing challenge in bioinformatics. In the process of predicting protein 3D structures, it is common that parts of an experimental structure are missing or parts of a predicted structure need to be remodeled. The process of predicting local protein structures of particular regions is called loop modeling. In this paper, five new loop modeling methods based on machine learning techniques, called NearLooper, ConLooper, ResLooper, HyLooper1, and HyLooper2 are proposed. NearLooper is based on the nearest neighbor technique. ConLooper applies deep convolutional neural networks to predict ${\mathrm{C}}_{{{\alpha }}}$Cα atoms distance matrix as an orientation-independent representation of protein structure. ResLooper uses residual neural networks instead of deep convolutional neural networks. HyLooper1 combines the results of NearLooper and ConLooper while HyLooper2 combines NearLooper and ResLooper. Three commonly used benchmarks for loop modeling are used to compare the performance between these methods and existing state-of-the-art methods. The experiment results show promising performance in which our best method improves existing state-of-the-art methods by 28 and 54 percent of average RMSD on two datasets while being comparable on the other one.

Improved protein structure reconstruction using secondary structures, contacts at higher distance thresholds, and non-contacts

BackgroundResidue-residue contacts are key features for accurate de novo protein structure prediction. For the optimal utilization of these predicted contacts in folding proteins accurately, it is important to study the challenges of reconstructing protein structures using true contacts. Because contact-guided protein modeling approach is valuable for predicting the folds of proteins that do not have structural templates, it is necessary for reconstruction studies to focus on hard-to-predict protein structures.ResultsUsing a data set consisting of 496 structural domains released in recent CASP experiments and a dataset of 150 representative protein structures, in this work, we discuss three techniques to improve the reconstruction accuracy using true contacts – adding secondary structures, increasing contact distance thresholds, and adding non-contacts. We find that reconstruction using secondary structures and contacts can deliver accuracy higher than using full contact maps. Similarly, we demonstrate that non-contacts can improve reconstruction accuracy not only when the used non-contacts are true but also when they are predicted. On the dataset consisting of 150 proteins, we find that by simply using low ranked predicted contacts as non-contacts and adding them as additional restraints, can increase the reconstruction accuracy by 5% when the reconstructed models are evaluated using TM-score.ConclusionsOur findings suggest that secondary structures are invaluable companions of contacts for accurate reconstruction. Confirming some earlier findings, we also find that larger distance thresholds are useful for folding many protein structures which cannot be folded using the standard definition of contacts. Our findings also suggest that for more accurate reconstruction using predicted contacts it is useful to predict contacts at higher distance thresholds (beyond 8 Å) and predict non-contacts.

Prediction of protein function using a deep convolutional neural network ensemble

BackgroundThe availability of large databases containing high resolution three-dimensional (3D) models of proteins in conjunction with functional annotation allows the exploitation of advanced supervised machine learning techniques for automatic protein function prediction.MethodsIn this work, novel shape features are extracted representing protein structure in the form of local (per amino acid) distribution of angles and amino acid distances, respectively. Each of the multi-channel feature maps is introduced into a deep convolutional neural network (CNN) for function prediction and the outputs are fused through support vector machines or a correlation-based k-nearest neighbor classifier. Two different architectures are investigated employing either one CNN per multi-channel feature set, or one CNN per image channel.ResultsCross validation experiments on single-functional enzymes (n = 44,661) from the PDB database achieved 90.1% correct classification, demonstrating an improvement over previous results on the same dataset when sequence similarity was not considered.DiscussionThe automatic prediction of protein function can provide quick annotations on extensive datasets opening the path for relevant applications, such as pharmacological target identification. The proposed method shows promise for structure-based protein function prediction, but sufficient data may not yet be available to properly assess the method’s performance on non-homologous proteins and thus reduce the confounding factor of evolutionary relationships.

A critical assessment of finite element modeling approach for protein dynamics.

Finite element (FE) modeling approach has emerged as an efficient way of calculating the dynamic properties of supramolecular protein structures and their complexes. Its efficiency mainly stems from the fact that the complexity of three-dimensional shape of a molecular surface dominates the computational cost rather than the molecular size or the number of atoms. However, no critical evaluation of the method has been made yet particularly for its sensitivity to the parameters used in model construction. Here, we make a close investigation on the effect of FE model parameters by analyzing 135 representative protein structures whose normal modes calculated using all-atom normal mode analysis are publicly accessible online. Results demonstrate that it is more beneficial to use a contour surface of electron densities as the molecular surface, in general, rather than to employ a solvent excluded surface, and that the solution accuracy is almost insensitive to the model parameters unless we avoid extreme values leading to an inaccurate depiction of the characteristic shapes.

Intelligent molecular orientation transformation in 3D printing system

Proper orientation of a molecular structure in three-dimensional (3D) printing could increase successful printing rates, reduce the amount of supporting material required, and shorten printing time. In traditional approaches, manual adjustment of a target object for its optimal orientation is inefficient and inconsistent and often requires several trials. Hence, manually and visually rotated results for molecular protein structures with complex, intertwined, rugged, and asymmetric surfaces are not satisfactory. In this study, we apply a grid-based principal component analysis (GPCA) method for an automatic object orientation prior to the physical printing stage. First, a down-sampled three-dimensional protein structure is constructed by lattice-space simulations, and then the proposed GPCA technology is applied to identify possible plane candidates with the largest projection area. Second, a vertical flipping operation is performed and evaluated for a smaller buried volume. Finally, the orientation of the rotated object is iteratively inspected and modified with subtle angle changes in order to further reduce the required supporting material. Several testing cases were used to illustrate the superior performance of the proposed algorithm. Specifically, 140 representative protein structures categorized into seven different groups were selected from the well-known Structural Classification of Proteins—extended database. As the results show, the protein structures were theoretically and heuristically rotated to their optimal orientations for the corresponding 3D printings. The proposed automatic orientation procedure could reduce 38.15% of the required supporting material on average. Furthermore, the expected printing time could be reduced by an average of 17.2 min for small-scale protein structures.

All-Atom Four-Body Knowledge-Based Statistical Potentials to Distinguish Native Protein Structures from Nonnative Folds.

Recent advances in understanding protein folding have benefitted from coarse-grained representations of protein structures. Empirical energy functions derived from these techniques occasionally succeed in distinguishing native structures from their corresponding ensembles of nonnative folds or decoys which display varying degrees of structural dissimilarity to the native proteins. Here we utilized atomic coordinates of single protein chains, comprising a large diverse training set, to develop and evaluate twelve all-atom four-body statistical potentials obtained by exploring alternative values for a pair of inherent parameters. Delaunay tessellation was performed on the atomic coordinates of each protein to objectively identify all quadruplets of interacting atoms, and atomic potentials were generated via statistical analysis of the data and implementation of the inverted Boltzmann principle. Our potentials were evaluated using benchmarking datasets from Decoys-‘R'-Us, and comparisons were made with twelve other physics- and knowledge-based potentials. Ranking 3rd, our best potential tied CHARMM19 and surpassed AMBER force field potentials. We illustrate how a generalized version of our potential can be used to empirically calculate binding energies for target-ligand complexes, using HIV-1 protease-inhibitor complexes for a practical application. The combined results suggest an accurate and efficient atomic four-body statistical potential for protein structure prediction and assessment.

An efficient algorithm for protein structure comparison using elastic shape analysis.

BackgroundProtein structure comparison play important role in in silico functional prediction of a new protein. It is also used for understanding the evolutionary relationships among proteins. A variety of methods have been proposed in literature for comparing protein structures but they have their own limitations in terms of accuracy and complexity with respect to computational time and space. There is a need to improve the computational complexity in comparison/alignment of proteins through incorporation of important biological and structural properties in the existing techniques.ResultsAn efficient algorithm has been developed for comparing protein structures using elastic shape analysis in which the sequence of 3D coordinates atoms of protein structures supplemented by additional auxiliary information from side-chain properties are incorporated. The protein structure is represented by a special function called square-root velocity function. Furthermore, singular value decomposition and dynamic programming have been employed for optimal rotation and optimal matching of the proteins, respectively. Also, geodesic distance has been calculated and used as the dissimilarity score between two protein structures. The performance of the developed algorithm is tested and found to be more efficient, i.e., running time reduced by 80–90 % without compromising accuracy of comparison when compared with the existing methods. Source codes for different functions have been developed in R. Also, user friendly web-based application called ProtSComp has been developed using above algorithm for comparing protein 3D structures and is accessible free.ConclusionsThe methodology and algorithm developed in this study is taking considerably less computational time without loss of accuracy (Table 2). The proposed algorithm is considering different criteria of representing protein structures using 3D coordinates of atoms and inclusion of residue wise molecular properties as auxiliary information.