Restraint validation of biomolecular structures determined by NMR in the Protein Data Bank.

Biomolecular structure analysis from experimental NMR studies generally relies on restraints derived from a combination of experimental and knowledge-based data. A challenge for the structural biology community has been a lack of standards for representing these restraints, preventing the establishment of uniform methods of model-vs-data structure validation against restraints and limiting interoperability between restraint-based structure modeling programs. The NEF and NMR-STAR formats provide a standardized approach for representing commonly used NMR restraints. Using these restraint formats, a standardized validation system for assessing structural models of biopolymers against restraints has been developed and implemented in the wwPDB OneDep data deposition-validation-biocuration system. The resulting wwPDB restraint violation report provides a model vs. data assessment of biomolecule structures determined using distance and dihedral restraints, with extensions to other restraint types currently being implemented. These tools are useful for assessing NMR models, as well as for assessing biomolecular structure predictions based on distance restraints.

Restraint Validation of Biomolecular Structures Determined by NMR in the Protein Data Bank.

Biomolecular structure analysis from experimental NMR studies generally relies on restraints derived from a combination of experimental and knowledge-based data. A challenge for the structural biology community has been a lack of standards for representing these restraints, preventing the establishment of uniform methods of model-vs-data structure validation against restraints and limiting interoperability between restraint-based structure modeling programs. The NMR exchange (NEF) and NMR-STAR formats provide a standardized approach for representing commonly used NMR restraints. Using these restraint formats, a standardized validation system for assessing structural models of biopolymers against restraints has been developed and implemented in the wwPDB OneDep data deposition-validation-biocuration system. The resulting wwPDB Restraint Violation Report provides a model vs. data assessment of biomolecule structures determined using distance and dihedral restraints, with extensions to other restraint types currently being implemented. These tools are useful for assessing NMR models, as well as for assessing biomolecular structure predictions based on distance restraints.

Biological Magnetic Resonance Data Bank.

The Biological Magnetic Resonance Data Bank (BMRB, https://bmrb.io) is the international open data repository for biomolecular nuclear magnetic resonance (NMR) data. Comprised of both empirical and derived data, BMRB has applications in the study of biomacromolecular structure and dynamics, biomolecular interactions, drug discovery, intrinsically disordered proteins, natural products, biomarkers, and metabolomics. Advances including GHz-class NMR instruments, national and trans-national NMR cyberinfrastructure, hybrid structural biology methods and machine learning are driving increases in the amount, type, and applications of NMR data in the biosciences. BMRB is a Core Archive and member of the World-wide Protein Data Bank (wwPDB).

Anomalous amide proton chemical shifts as signatures of hydrogen bonding to aromatic sidechains.

Hydrogen bonding between an amide group and the p-π cloud of an aromatic ring was first identified in a protein in the 1980s. Subsequent surveys of high-resolution X-ray crystal structures found multiple instances, but their preponderance was determined to be infrequent. Hydrogen atoms participating in a hydrogen bond to the p-π cloud of an aromatic ring are expected to experience an upfield chemical shift arising from a shielding ring current shift. We surveyed the Biological Magnetic Resonance Data Bank for amide hydrogens exhibiting unusual shifts as well as corroborating nuclear Overhauser effects between the amide protons and ring protons. We found evidence that Trp residues are more likely to be involved in p-π hydrogen bonds than other aromatic amino acids, whereas His residues are more likely to be involved in in-plane hydrogen bonds, with a ring nitrogen acting as the hydrogen acceptor. The p-π hydrogen bonds may be more abundant than previously believed. The inclusion in NMR structure refinement protocols of shift effects in amide protons from aromatic sidechains, or explicit hydrogen bond restraints between amides and aromatic rings, could improve the local accuracy of sidechain orientations in solution NMR protein structures, but their impact on global accuracy is likely be limited.

Merging NMR Data and Computation Facilitates Data-Centered Research.

The Biological Magnetic Resonance Data Bank (BMRB) has served the NMR structural biology community for 40 years, and has been instrumental in the development of many widely-used tools. It fosters the reuse of data resources in structural biology by embodying the FAIR data principles (Findable, Accessible, Inter-operable, and Re-usable). NMRbox is less than a decade old, but complements BMRB by providing NMR software and high-performance computing resources, facilitating the reuse of software resources. BMRB and NMRbox both facilitate reproducible research. NMRbox also fosters the development and deployment of complex meta-software. Combining BMRB and NMRbox helps speed and simplify workflows that utilize BMRB, and enables facile federation of BMRB with other data repositories. Utilization of BMRB and NMRbox in tandem will enable additional advances, such as machine learning, that are poised to become increasingly powerful.

BioMagResBank (BMRB) as a Resource for Structural Biology.

The Biological Magnetic Resonance Data Bank (BioMagResBank or BMRB), founded in 1988, serves as the archive for data generated by nuclear magnetic resonance (NMR) spectroscopy of biological systems. NMR spectroscopy is unique among biophysical approaches in its ability to provide a broad range of atomic and higher-level information relevant to the structural, dynamic, and chemical properties of biological macromolecules, as well as report on metabolite and natural product concentrations in complex mixtures and their chemical structures. BMRB became a core member of the Worldwide Protein Data Bank (wwPDB) in 2007, and the BMRB archive is now a core archive of the wwPDB. Currently, about 10% of the structures deposited into the PDB archive are based on NMR spectroscopy. BMRB stores experimental and derived data from biomolecular NMR studies. Newer BMRB biopolymer depositions are divided about evenly between those associated with structure determinations (atomic coordinates and supporting information archived in the PDB) and those reporting experimental information on molecular dynamics, conformational transitions, ligand binding, assigned chemical shifts, or other results from NMR spectroscopy. BMRB also provides resources for NMR studies of metabolites and other small molecules that are often macromolecular ligands and/or nonstandard residues. This chapter is directed to the structural biology community rather than the metabolomics and natural products community. Our goal is to describe various BMRB services offered to structural biology researchers and how they can be accessed and utilized. These services can be classified into four main groups: (1) data deposition, (2) data retrieval, (3) data analysis, and (4) services for NMR spectroscopists and software developers. The chapter also describes the NMR-STAR data format used by BMRB and the tools provided to facilitate its use. For programmers, BMRB offers an application programming interface (API) and libraries in the Python and R languages that enable users to develop their own BMRB-based tools for data analysis, visualization, and manipulation of NMR-STAR formatted files. BMRB also provides users with direct access tools through the NMRbox platform.

Structure determination using solution NMR: Is it worth the effort?

It has been almost 40 years since solution NMR joined X-ray crystallography as a technique for determining high-resolution structures of proteins. Since then NMR derived structure has contributed in fundamental ways to our understanding of the function of biomolecules. With the already existing mature field of X-ray crystallography and the emergence of cryo-EM as techniques to tackle high-resolution structures of large protein complexes, the role of NMR in structure determination has been questioned. However, NMR has the unique ability to recapitulate the dynamic motion of proteins in their structures, while size limitations of the biomolecular systems that can be routinely studied still present challenges. The field has continually developed methodology and instrumentation since its introduction, pushing its frontiers and redefining its limits. Here we present a brief overview of NMR-based structure determination over the past 40 years. We outline the current state of the field and look ahead to the challenges that still need to be addressed to realize the future potential of NMR as a structural technique.

NMR-STAR: comprehensive ontology for representing, archiving and exchanging data from nuclear magnetic resonance spectroscopic experiments.

The growth of the biological nuclear magnetic resonance (NMR) field and the development of new experimental technology have mandated the revision and enlargement of the NMR-STAR ontology used to represent experiments, spectral and derived data, and supporting metadata. We present here a brief description of the NMR-STAR ontology and software tools for manipulating NMR-STAR data files, editing the files, extracting selected data, and creating data visualizations. Detailed information on these is accessible from the links provided.

Worldwide Protein Data Bank biocuration supporting open access to high-quality 3D structural biology data.

Database URL: https://www.wwpdb.org/.

Validation of Structures in the Protein Data Bank.

The Worldwide PDB recently launched a deposition, biocuration, and validation tool: OneDep. At various stages of OneDep data processing, validation reports for three-dimensional structures of biological macromolecules are produced. These reports are based on recommendations of expert task forces representing crystallography, nuclear magnetic resonance, and cryoelectron microscopy communities. The reports provide useful metrics with which depositors can evaluate the quality of the experimental data, the structural model, and the fit between them. The validation module is also available as a stand-alone web server and as a programmatically accessible web service. A growing number of journals require the official wwPDB validation reports (produced at biocuration) to accompany manuscripts describing macromolecular structures. Upon public release of the structure, the validation report becomes part of the public PDB archive. Geometric quality scores for proteins in the PDB archive have improved over the past decade.

Increasing rigor in NMR-based metabolomics through validated and open source tools.

The metabolome, the collection of small molecules associated with an organism, is a growing subject of inquiry, with the data utilized for data-intensive systems biology, disease diagnostics, biomarker discovery, and the broader characterization of small molecules in mixtures. Owing to their close proximity to the functional endpoints that govern an organism's phenotype, metabolites are highly informative about functional states. The field of metabolomics identifies and quantifies endogenous and exogenous metabolites in biological samples. Information acquired from nuclear magnetic spectroscopy (NMR), mass spectrometry (MS), and the published literature, as processed by statistical approaches, are driving increasingly wider applications of metabolomics. This review focuses on the role of databases and software tools in advancing the rigor, robustness, reproducibility, and validation of metabolomics studies.

Understanding the fabric of protein crystals: computational classification of biological interfaces and crystal contacts.

Modern structural biology still draws the vast majority of information from crystallography, a technique where the objects being investigated are embedded in a crystal lattice. Given the complexity and variety of those objects, it becomes fundamental to computationally assess which of the interfaces in the lattice are biologically relevant and which are simply crystal contacts. Since the mid-1990s, several approaches have been applied to obtain high-accuracy classification of crystal contacts and biological protein-protein interfaces. This review provides an overview of the concepts and main approaches to protein interface classification: thermodynamic estimation of interface stability, evolutionary approaches based on conservation of interface residues, and co-occurrence of the interface across different crystal forms. Among the three categories, evolutionary approaches offer the strongest promise for improvement, thanks to the incessant growth in sequence knowledge. Importantly, protein interface classification algorithms can also be used on multimeric structures obtained using other high-resolution techniques or for protein assembly design or validation purposes. A key issue linked to protein interface classification is the identification of the biological assembly of a crystal structure and the analysis of its symmetry. Here, we highlight the most important concepts and problems to be overcome in assembly prediction. Over the next few years, tools and concepts of interface classification will probably become more frequently used and integrated in several areas of structural biology and structural bioinformatics. Among the main challenges for the future are better addressing of weak interfaces and the application of interface classification concepts to prediction problems like protein-protein docking.

A PDB-wide, evolution-based assessment of protein-protein interfaces.

BACKGROUND: Thanks to the growth in sequence and structure databases, more than 50 million sequences are now available in UniProt and 100,000 structures in the PDB. Rich information about protein-protein interfaces can be obtained by a comprehensive study of protein contacts in the PDB, their sequence conservation and geometric features. RESULTS: An automated computational pipeline was developed to run our Evolutionary Protein-Protein Interface Classifier (EPPIC) software on the entire PDB and store the results in a relational database, currently containing > 800,000 interfaces. This allows the analysis of interface data on a PDB-wide scale. Two large benchmark datasets of biological interfaces and crystal contacts, each containing about 3000 entries, were automatically generated based on criteria thought to be strong indicators of interface type. The BioMany set of biological interfaces includes NMR dimers solved as crystal structures and interfaces that are preserved across diverse crystal forms, as catalogued by the Protein Common Interface Database (ProtCID) from Xu and Dunbrack. The second dataset, XtalMany, is derived from interfaces that would lead to infinite assemblies and are therefore crystal contacts. BioMany and XtalMany were used to benchmark the EPPIC approach. The performance of EPPIC was also compared to classifications from the Protein Interfaces, Surfaces, and Assemblies (PISA) program on a PDB-wide scale, finding that the two approaches give the same call in about 88% of PDB interfaces. By comparing our safest predictions to the PDB author annotations, we provide a lower-bound estimate of the error rate of biological unit annotations in the PDB. Additionally, we developed a PyMOL plugin for direct download and easy visualization of EPPIC interfaces for any PDB entry. Both the datasets and the PyMOL plugin are available at http://www.eppic-web.org/ewui/\#downloads. CONCLUSIONS: Our computational pipeline allows us to analyze protein-protein contacts and their sequence conservation across the entire PDB. Two new benchmark datasets are provided, which are over an order of magnitude larger than existing manually curated ones. These tools enable the comprehensive study of several aspects of protein-protein contacts in the PDB and represent a basis for future, even larger scale studies of protein-protein interactions.

An analysis of oligomerization interfaces in transmembrane proteins.

BACKGROUND: The amount of transmembrane protein (TM) structures solved to date is now large enough to attempt large scale analyses. In particular, extensive studies of oligomeric interfaces in the transmembrane region are now possible. RESULTS: We have compiled the first fully comprehensive set of validated transmembrane protein interfaces in order to study their features and assess what differentiates them from their soluble counterparts. CONCLUSIONS: The general features of TM interfaces do not differ much from those of soluble proteins: they are large, tightly packed and possess many interface core residues. In our set, membrane lipids were not found to significantly mediate protein-protein interfaces. Although no G protein-coupled receptor (GPCR) was included in the validated set, we analyzed the crystallographic dimerization interfaces proposed in the literature. We found that the putative dimer interfaces proposed for class A GPCRs do not show the usual patterns of stable biological interfaces, neither in terms of evolution nor of packing, thus they likely correspond to crystal interfaces. We cannot however rule out the possibility that they constitute transient or weak interfaces. In contrast we do observe a clear signature of biological interface for the proposed dimer of the class F human Smoothened receptor.

Mapping of protein structural ensembles by chemical shifts.

Applying the chemical shift prediction programs SHIFTX and SHIFTS to a data base of protein structures with known chemical shifts we show that the averaged chemical shifts predicted from the structural ensembles explain better the experimental data than the lowest energy structures. This is in agreement with the fact that proteins in solution occur in multiple conformational states in fast exchange on the chemical shift time scale. However, in contrast to the real conditions in solution at ambient temperatures, the standard NMR structural calculation methods as well chemical shift prediction methods are optimized to predict the lowest energy ground state structure that is only weakly populated at physiological temperatures. An analysis of the data shows that a chemical shift prediction can be used as measure to define the minimum size of the structural bundle required for a faithful description of the structural ensemble.

Protein structure calculation with data imputation: the use of substitute restraints.

The amount of experimental restraints e.g., NOEs is often too small for calculating high quality three-dimensional structures by restrained molecular dynamics. Considering this as a typical missing value problem we propose here a model based data imputation technique that should lead to an improved estimation of the correct structure. The novel automated method implemented in AUREMOL makes a more efficient use of the experimental information to obtain NMR structures with higher accuracy. It creates a large set of substitute restraints that are used either alone or together with the experimental restraints. The new approach was successfully tested on three examples: firstly, the Ras-binding domain of Byr2 from Schizosaccharomyces pombe, the mutant HPr (H15A) from Staphylococcus aureus, and a X-ray structure of human ubiquitin. In all three examples, the quality of the resulting final bundles was improved considerably by the use of additional substitute restraints, as assessed quantitatively by the calculation of RMSD values to the "true" structure and NMR R-factors directly calculated from the original NOESY spectra or the published diffraction data.

Chemical shift optimization in multidimensional NMR spectra by AUREMOL-SHIFTOPT.

A problem often encountered in multidimensional NMR-spectroscopy is that an existing chemical shift list of a protein has to be used to assign an experimental spectrum but does not fit sufficiently well for a safe assignment. A similar problem occurs when temperature or pressure series of n-dimensional spectra are to be evaluated automatically. We have developed two different algorithms, AUREMOL-SHIFTOPT1 and AUREMOL-SHIFTOPT2 that fulfill this task. In the present contribution their performance is analyzed employing a set of simulated and experimental two-dimensional and three-dimensional spectra obtained from three different proteins. A new z-score based on atom and amino acid specific chemical shift distributions is introduced to weight the chemical shift contributions in different dimensions properly.