Knowledge-based Statistical Potentials Research Articles

LociPARSE: a locality-aware invariant point attention model for scoring RNA 3D structures.

A scoring function that can reliably assess the accuracy of a 3D RNA structural model in the absence of experimental structure is not only important for model evaluation and selection but also useful for scoring-guided conformational sampling. However, high-fidelity RNA scoring has proven to be difficult using conventional knowledge-based statistical potentials and currently-available machine learning-based approaches. Here we present lociPARSE, a locality-aware invariant point attention architecture for scoring RNA 3D structures. Unlike existing machine learning methods that estimate superposition-based root mean square deviation (RMSD), lociPARSE estimates Local Distance Difference Test (lDDT) scores capturing the accuracy of each nucleotide and its surrounding local atomic environment in a superposition-free manner, before aggregating information to predict global structural accuracy. Tested on multiple datasets including CASP15, lociPARSE significantly outperforms existing statistical potentials (rsRNASP, cgRNASP, DFIRE-RNA, and RASP) and machine learning methods (ARES and RNA3DCNN) across complementary assessment metrics. lociPARSE is freely available at https://github.com/Bhattacharya-Lab/lociPARSE.

CgRNASP: coarse-grained statistical potentials with residue separation for RNA structure evaluation.

Knowledge-based statistical potentials are very important for RNA 3-dimensional (3D) structure prediction and evaluation. In recent years, various coarse-grained (CG) and all-atom models have been developed for predicting RNA 3D structures, while there is still lack of reliable CG statistical potentials not only for CG structure evaluation but also for all-atom structure evaluation at high efficiency. In this work, we have developed a series of residue-separation-based CG statistical potentials at different CG levels for RNA 3D structure evaluation, namely cgRNASP, which is composed of long-ranged and short-ranged interactions by residue separation. Compared with the newly developed all-atom rsRNASP, the short-ranged interaction in cgRNASP was involved more subtly and completely. Our examinations show that, the performance of cgRNASP varies with CG levels and compared with rsRNASP, cgRNASP has similarly good performance for extensive types of test datasets and can have slightly better performance for the realistic dataset-RNA-Puzzles dataset. Furthermore, cgRNASP is strikingly more efficient than all-atom statistical potentials/scoring functions, and can be apparently superior to other all-atom statistical potentials and scoring functions trained from neural networks for the RNA-Puzzles dataset. cgRNASP is available at https://github.com/Tan-group/cgRNASP.

RsRNASP: A residue-separation-based statistical potential for RNA 3D structure evaluation

Knowledge-based statistical potentials have been shown to be rather effective in protein 3-dimensional (3D) structure evaluation and prediction. Recently, several statistical potentials have been developed for RNA 3D structure evaluation, while their performances are either still at a low level for the test datasets from structure prediction models or dependent on the “black-box” process through neural networks. In this work, we have developed an all-atom distance-dependent statistical potential based on residue separation for RNA 3D structure evaluation, namely rsRNASP, which is composed of short- and long-ranged potentials distinguished by residue separation. The extensive examinations against available RNA test datasets show that rsRNASP has apparently higher performance than the existing statistical potentials for the realistic test datasets with large RNAs from structure prediction models, including the newly released RNA-Puzzles dataset, and is comparable to the existing top statistical potentials for the test datasets with small RNAs or near-native decoys. In addition, rsRNASP is superior to RNA3DCNN, a recently developed scoring function through 3D convolutional neural networks. rsRNASP and the relevant databases are available to the public.

DISPOT: a simple knowledge-based protein domain interaction statistical potential.

MotivationThe complexity of protein–protein interactions (PPIs) is further compounded by the fact that an average protein consists of two or more domains, structurally and evolutionary independent subunits. Experimental studies have demonstrated that an interaction between a pair of proteins is not carried out by all domains constituting each protein, but rather by a select subset. However, determining which domains from each protein mediate the corresponding PPI is a challenging task.ResultsHere, we present domain interaction statistical potential (DISPOT), a simple knowledge-based statistical potential that estimates the propensity of an interaction between a pair of protein domains, given their structural classification of protein (SCOP) family annotations. The statistical potential is derived based on the analysis of >352 000 structurally resolved PPIs obtained from DOMMINO, a comprehensive database of structurally resolved macromolecular interactions.Availability and implementationDISPOT is implemented in Python 2.7 and packaged as an open-source tool. DISPOT is implemented in two modes, basic and auto-extraction. The source code for both modes is available on GitHub: https://github.com/korkinlab/dispot and standalone docker images on DockerHub: https://hub.docker.com/r/korkinlab/dispot. The web server is freely available at http://dispot.korkinlab.org/.Supplementary information Supplementary data are available at Bioinformatics online.

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

BackgroundThe knowledge-based statistical potential has been widely used in protein structure modeling and model quality assessment. They are commonly evaluated based on their abilities of native recognition as well as decoy discrimination. However, these two aspects are found to be mutually exclusive in many statistical potentials.ResultsWe developed an atomic ANgle- and DIStance-dependent (ANDIS) statistical potential for protein structure quality assessment with distance cutoff being a tunable parameter. When distance cutoff is ≤9.0 Å, “effective atomic interaction” is employed to enhance the ability of native recognition. For a distance cutoff of ≥10 Å, the distance-dependent atom-pair potential with random-walk reference state is combined to strengthen the ability of decoy discrimination. Benchmark tests on 632 structural decoy sets from diverse sources demonstrate that ANDIS outperforms other state-of-the-art potentials in both native recognition and decoy discrimination.ConclusionsDistance cutoff is a crucial parameter for distance-dependent statistical potentials. A lower distance cutoff is better for native recognition, while a higher one is favorable for decoy discrimination. The ANDIS potential is freely available as a standalone application at http://qbp.hzau.edu.cn/ANDIS/.

What is the best reference state for building statistical potentials in RNA 3D structure evaluation?

Knowledge-based statistical potentials have been shown to be efficient in protein structure evaluation/prediction, and the core difference between various statistical potentials is attributed to the choice of reference states. However, for RNA 3D structure evaluation, a comprehensive examination on reference states is still lacking. In this work, we built six statistical potentials based on six reference states widely used in protein structure evaluation, including averaging, quasi-chemical approximation, atom-shuffled, finite-ideal-gas, spherical-noninteracting, and random-walk-chain reference states, and we examined the six reference states against three RNA test sets including six subsets. Our extensive examinations show that, overall, for identifying native structures and ranking decoy structures, the finite-ideal-gas and random-walk-chain reference states are slightly superior to others, while for identifying near-native structures, there is only a slight difference between these reference states. Our further analyses show that the performance of a statistical potential is apparently dependent on the quality of the training set. Furthermore, we found that the performance of a statistical potential is closely related to the origin of test sets, and for the three realistic test subsets, the six statistical potentials have overall unsatisfactory performance. This work presents a comprehensive examination on the existing reference states and statistical potentials for RNA 3D structure evaluation.

RNA3DCNN: Local and global quality assessments of RNA 3D structures using 3D deep convolutional neural networks.

Quality assessment is essential for the computational prediction and design of RNA tertiary structures. To date, several knowledge-based statistical potentials have been proposed and proved to be effective in identifying native and near-native RNA structures. All these potentials are based on the inverse Boltzmann formula, while differing in the choice of the geometrical descriptor, reference state, and training dataset. Via an approach that diverges completely from the conventional statistical potentials, our work explored the power of a 3D convolutional neural network (CNN)-based approach as a quality evaluator for RNA 3D structures, which used a 3D grid representation of the structure as input without extracting features manually. The RNA structures were evaluated by examining each nucleotide, so our method can also provide local quality assessment. Two sets of training samples were built. The first one included 1 million samples generated by high-temperature molecular dynamics (MD) simulations and the second one included 1 million samples generated by Monte Carlo (MC) structure prediction. Both MD and MC procedures were performed for a non-redundant set of 414 RNAs. For two training datasets (one including only MD training samples and the other including both MD and MC training samples), we trained two neural networks, named RNA3DCNN_MD and RNA3DCNN_MDMC, respectively. The former is suitable for assessing near-native structures, while the latter is suitable for assessing structures covering large structural space. We tested the performance of our method and made comparisons with four other traditional scoring functions. On two of three test datasets, our method performed similarly to the state-of-the-art traditional scoring function, and on the third test dataset, our method was far superior to other scoring functions. Our method can be downloaded from https://github.com/lijunRNA/RNA3DCNN.

All-atom four-body knowledge-based statistical potential to distinguish native tertiary RNA structures from nonnative folds

Scientific breakthroughs in recent decades have uncovered the capability of RNA molecules to fulfill a wide array of structural, functional, and regulatory roles in living cells, leading to a concomitantly significant increase in both the number and diversity of experimentally determined RNA three-dimensional (3D) structures. Atomic coordinates from a representative training set of solved RNA structures, displaying low sequence and structure similarity, facilitate derivation of knowledge-based energy functions. Here we develop an all-atom four-body statistical potential and evaluate its capacity to distinguish native RNA 3D structures from nonnative folds based on calculated free energy scores. Atomic four-body nearest-neighbors are objectively identified by their occurrence as tetrahedral vertices in the Delaunay tessellations of RNA structures, and rates of atomic quadruplet interactions expected by chance are obtained from a multinomial reference distribution. Our four-body energy function, referred to as RAMP (ribonucleic acids multibody potential), is subsequently derived by applying the inverted Boltzmann principle to the frequency data, yielding an energy score for each type of atomic quadruplet interaction. Several well-known benchmark datasets reveal that RAMP is comparable with, and often outperforms, existing knowledge- and physics-based energy functions. To the best of our knowledge, this is the first study detailing an RNA tertiary structure-based multibody statistical potential and its comparative evaluation.

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

[This corrects the article DOI: 10.1155/2017/5760612.].

Alchemical Free Energy Calculations for Nucleotide Mutations in Protein-DNA Complexes.

Nucleotide-sequence-dependent interactions between proteins and DNA are responsible for a wide range of gene regulatory functions. Accurate and generalizable methods to evaluate the strength of protein-DNA binding have long been sought. While numerous computational approaches have been developed, most of them require fitting parameters to experimental data to a certain degree, e.g., machine learning algorithms or knowledge-based statistical potentials. Molecular-dynamics-based free energy calculations offer a robust, system-independent, first-principles-based method to calculate free energy differences upon nucleotide mutation. We present an automated procedure to set up alchemical MD-based calculations to evaluate free energy changes occurring as the result of a nucleotide mutation in DNA. We used these methods to perform a large-scale mutation scan comprising 397 nucleotide mutation cases in 16 protein-DNA complexes. The obtained prediction accuracy reaches 5.6 kJ/mol average unsigned deviation from experiment with a correlation coefficient of 0.57 with respect to the experimentally measured free energies. Overall, the first-principles-based approach performed on par with the molecular modeling approaches Rosetta and FoldX. Subsequently, we utilized the MD-based free energy calculations to construct protein-DNA binding profiles for the zinc finger protein Zif268. The calculation results compare remarkably well with the experimentally determined binding profiles. The software automating the structure and topology setup for alchemical calculations is a part of the pmx package; the utilities have also been made available online at http://pmx.mpibpc.mpg.de/dna_webserver.html .

SURPASS Low-Resolution Coarse-Grained Protein Modeling.

Coarse-grained modeling of biomolecules has a very important role in molecular biology. In this work we present a novel SURPASS (Single United Residue per Pre-Averaged Secondary Structure fragment) model of proteins that can be an interesting alternative for existing coarse-grained models. The design of the model is unique and strongly supported by the statistical analysis of structural regularities characteristic for protein systems. Coarse-graining of protein chain structures assumes a single center of interactions per residue and accounts for preaveraged effects of four adjacent residue fragments. Knowledge-based statistical potentials encode complex interaction patterns of these fragments. Using the Replica Exchange Monte Carlo sampling scheme and a generic version of the SURPASS force field we performed test simulations of a representative set of single-domain globular proteins. The method samples a significant part of conformational space and reproduces protein structures, including native-like, with surprisingly good accuracy. Future extension of the SURPASS model on large biomacromolecular systems is briefly discussed.

VoroMQA: Assessment of protein structure quality using interatomic contact areas.

In the absence of experimentally determined protein structure many biological questions can be addressed using computational structural models. However, the utility of protein structural models depends on their quality. Therefore, the estimation of the quality of predicted structures is an important problem. One of the approaches to this problem is the use of knowledge-based statistical potentials. Such methods typically rely on the statistics of distances and angles of residue-residue or atom-atom interactions collected from experimentally determined structures. Here, we present VoroMQA (Voronoi tessellation-based Model Quality Assessment), a new method for the estimation of protein structure quality. Our method combines the idea of statistical potentials with the use of interatomic contact areas instead of distances. Contact areas, derived using Voronoi tessellation of protein structure, are used to describe and seamlessly integrate both explicit interactions between protein atoms and implicit interactions of protein atoms with solvent. VoroMQA produces scores at atomic, residue, and global levels, all in the fixed range from 0 to 1. The method was tested on the CASP data and compared to several other single-model quality assessment methods. VoroMQA showed strong performance in the recognition of the native structure and in the structural model selection tests, thus demonstrating the efficacy of interatomic contact areas in estimating protein structure quality. The software implementation of VoroMQA is freely available as a standalone application and as a web server at http://bioinformatics.lt/software/voromqa. Proteins 2017; 85:1131-1145. © 2017 Wiley Periodicals, Inc.

Using sequence signatures and kink-turn motifs in knowledge-based statistical potentials for RNA structure prediction

Kink turns are widely occurring motifs in RNA, located in internal loops and associated with many biological functions including translation, regulation and splicing. The associated sequence pattern, a 3-nt bulge and G-A, A-G base-pairs, generates an angle of ∼50° along the helical axis due to A-minor interactions. The conserved sequence and distinct secondary structures of kink-turns (k-turn) suggest computational folding rules to predict k-turn-like topologies from sequence. Here, we annotate observed k-turn motifs within a non-redundant RNA dataset based on sequence signatures and geometrical features, analyze bending and torsion angles, and determine distinct knowledge-based potentials with and without k-turn motifs. We apply these scoring potentials to our RAGTOP (RNA-As-Graph-Topologies) graph sampling protocol to construct and sample coarse-grained graph representations of RNAs from a given secondary structure. We present graph-sampling results for 35 RNAs, including 12 k-turn and 23 non k-turn internal loops, and compare the results to solved structures and to RAGTOP results without special k-turn potentials. Significant improvements are observed with the updated scoring potentials compared to the k-turn-free potentials. Because k-turns represent a classic example of sequence/structure motif, our study suggests that other such motifs with sequence signatures and unique geometrical features can similarly be utilized for RNA structure prediction and design.

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.

WT Human γD Crystallin Promotes Aggregation of its Oxidation-Mimicking Mutants

Non-native protein conformers generated by mutation or chemical damage template aggregation of wild-type, undamaged polypeptides in diseases ranging from amyotrophic lateral sclerosis to cancer. We tested for such interactions in the natively monomeric human eye lens protein γD-crystallin, whose aggregation leads to cataract disease. The oxidation-mimicking W42Q mutant of γD-crystallin formed non-native polymers starting from a native-like state under physiological conditions in vitro. Aggregation occurred in the temperature range 35-45 oC, in which the mutant protein began to lose the native conformation of its N-terminal domain. Surprisingly, it was wild-type γD-crystallin that promoted W42Q polymerization in a catalytic manner, even at mutant concentrations too low for homogeneous nucleation to occur. WT also downshifted the temperature range of W42Q aggregation. Aggregation required formation of a non-native internal disulfide bond in the mutant protein, which may result from a conformational rearrangement catalyzed by WT binding. Similar results were obtained for the W42R mutant, which causes congenital cataract. Monte-Carlo simulations with a knowledge-based statistical potential showed melting of specific residue contacts as a result of hydrophilic Trp substitutions. WT and W42Q populated similar early unfolding intermediates, which were able to domain-swap in two-chain simulations. The WT/W42Q heterodimer was longer-lived in silico than the W42Q/W42Q homodimer, suggesting a possible mechanism for WT-catalyzed aggregation of mutant or modified protein as oxidative damage accumulates during the course of aging.

3dRNAscore: a distance and torsion angle dependent evaluation function of 3D RNA structures.

Model evaluation is a necessary step for better prediction and design of 3D RNA structures. For proteins, this has been widely studied and the knowledge-based statistical potential has been proved to be one of effective ways to solve this problem. Currently, a few knowledge-based statistical potentials have also been proposed to evaluate predicted models of RNA tertiary structures. The benchmark tests showed that they can identify the native structures effectively but further improvements are needed to identify near-native structures and those with non-canonical base pairs. Here, we present a novel knowledge-based potential, 3dRNAscore, which combines distance-dependent and dihedral-dependent energies. The benchmarks on different testing datasets all show that 3dRNAscore are more efficient than existing evaluation methods in recognizing native state from a pool of near-native states of RNAs as well as in ranking near-native states of RNA models.

Computational prediction of riboswitch tertiary structures including pseudoknots by RAGTOP: a hierarchical graph sampling approach.

The modular organization of RNA structure has been exploited in various computational and theoretical approaches to identify RNA tertiary (3D) motifs and assemble RNA structures. Riboswitches exemplify this modularity in terms of both structural and functional adaptability of RNA components. Here, we extend our computational approach based on tree graph sampling to the prediction of riboswitch topologies by defining additional edges to mimick pseudoknots. Starting from a secondary (2D) structure, we construct an initial graph deduced from predicted junction topologies by our data-mining algorithm RNAJAG trained on known RNAs; we sample these graphs in 3D space guided by knowledge-based statistical potentials derived from bending and torsion measures of internal loops as well as radii of gyration for known RNAs. We present graph sampling results for 10 representative riboswitches, 6 of them with pseudoknots, and compare our predictions to solved structures based on global and local RMSD measures. Our results indicate that the helical arrangements in riboswitches can be approximated using our combination of modified 3D tree graph representations for pseudoknots, junction prediction, graph moves, and scoring functions. Future challenges in the field of riboswitch prediction and design are also discussed.

Carbohydrate-binding protein identification by coupling structural similarity searching with binding affinity prediction.

Carbohydrate-binding proteins (CBPs) are potential biomarkers and drug targets. However, the interactions between carbohydrates and proteins are challenging to study experimentally and computationally because of their low binding affinity, high flexibility, and the lack of a linear sequence in carbohydrates as exists in RNA, DNA, and proteins. Here, we describe a structure-based function-prediction technique called SPOT-Struc that identifies carbohydrate-recognizing proteins and their binding amino acid residues by structural alignment program SPalign and binding affinity scoring according to a knowledge-based statistical potential based on the distance-scaled finite-ideal gas reference state (DFIRE). The leave-one-out cross-validation of the method on 113 carbohydrate-binding domains and 3442 noncarbohydrate binding proteins yields a Matthews correlation coefficient of 0.56 for SPalign alone and 0.63 for SPOT-Struc (SPalign + binding affinity scoring) for CBP prediction. SPOT-Struc is a technique with high positive predictive value (79% correct predictions in all positive CBP predictions) with a reasonable sensitivity (52% positive predictions in all CBPs). The sensitivity of the method was changed slightly when applied to 31 APO (unbound) structures found in the protein databank (14/31 for APO versus 15/31 for HOLO). The result of SPOT-Struc will not change significantly if highly homologous templates were used. SPOT-Struc predicted 19 out of 2076 structural genome targets as CBPs. In particular, one uncharacterized protein in Bacillus subtilis (1oq1A) was matched to galectin-9 from Mus musculus. Thus, SPOT-Struc is useful for uncovering novel carbohydrate-binding proteins. SPOT-Struc is available at http://sparks-lab.org.

Protein Model Quality Assessment Prediction by using a Residue Specific Statistical Potential

A residue specific approach is introduced for enhancing the accuracy of knowledge-based statistical potentials for quality assessment of protein models. The proposed method relies on an assumption that common substructure motifs among different protein folds can still share similar patterns of interactions with neighboring residues. Instead of using the classic approach, in which a set of non-homologous whole-chain structures are used to derive a potential, we created a potential that was composed of independent sub-potentials; each of them was specific to a certain residue in the target sequence. To achieve that, the target sequence was split into short linear segments, then every segment was threaded through PDB to find proteins that share sequence similarity with this segment. Then, after removal the hits that were homologous to the target sequence, the remaining proteins were used to derive a statistical sub-potential for the residue that was in the center of a certain segment that had been threaded through PDB. This procedure was performed for every residue in the target protein. We applied this methodology to create a residue-specific variant of the DFIRE statistical potential. For CASP9 single-domain targets, the average Pearson's correlation coefficient per target between the pseudo-energy predicted by the residue-specific DIFIRE potential and the GDT_TS score was 0.656. The classic DFIRE potential achieved the correlation coefficient of 0.561. We believe that, for the current size (which is still growing) of the PDB database, this methodology can be successfully applied to increase the accuracy of other state-of-the-art statistical potentials.

Membrane protein orientation and refinement using a knowledge-based statistical potential.

BackgroundRecent increases in the number of deposited membrane protein crystal structures necessitate the use of automated computational tools to position them within the lipid bilayer. Identifying the correct orientation allows us to study the complex relationship between sequence, structure and the lipid environment, which is otherwise challenging to investigate using experimental techniques due to the difficulty in crystallising membrane proteins embedded within intact membranes.ResultsWe have developed a knowledge-based membrane potential, calculated by the statistical analysis of transmembrane protein structures, coupled with a combination of genetic and direct search algorithms, and demonstrate its use in positioning proteins in membranes, refinement of membrane protein models and in decoy discrimination.ConclusionsOur method is able to quickly and accurately orientate both alpha-helical and beta-barrel membrane proteins within the lipid bilayer, showing closer agreement with experimentally determined values than existing approaches. We also demonstrate both consistent and significant refinement of membrane protein models and the effective discrimination between native and decoy structures. Source code is available under an open source license from http://bioinf.cs.ucl.ac.uk/downloads/memembed/.