An Improved Infrastructure for Privacy-Preserving Analysis of Patient Data.

Incorporating healthcare data from different sources is crucial for a better understanding of patient (sub)populations. However, data centralization raises concerns about data privacy and governance. In this work, we present an improved infrastructure that allows privacy-preserving analysis of patient data: vantage6 v3. For this new version, we describe its architecture and upgraded functionality, which allows algorithms running at each party to communicate with one another through a virtual private network (while still being isolated from the public internet to reduce the risk of data leakage). This allows the execution of different types of algorithms (e.g., multi-party computation) that were practically infeasible before, as showcased by the included examples. The (continuous) development of this type of infrastructure is fundamental to meet the current and future demands of healthcare research with a strong emphasis on preserving the privacy of sensitive patient data.

LAHMA: structure analysis through local annotation of homology-matched amino acids.

Comparison of homologous structure models is a key step in analyzing protein structure. With a wealth of homologous structures, comparison becomes a tedious process, and often only a small (user-biased) selection of data is used. A multitude of structural superposition algorithms are then typically used to visualize the structures together in 3D and to compare them. Here, the Local Annotation of Homology-Matched Amino acids (LAHMA) website (https://lahma.pdb-redo.eu) is presented, which compares any structure model with all of its close homologs from the PDB-REDO databank. LAHMA displays structural features in sequence space, allowing users to uncover differences between homologous structure models that can be analyzed for their relevance to chemistry or biology. LAHMA visualizes numerous structural features, also allowing one-click comparison of structure-quality plots (for example the Ramachandran plot) and `in-browser' structural visualization of 3D models.

Building and rebuilding N-glycans in protein structure models.

N-Glycosylation is one of the most common post-translational modifications and is implicated in, for example, protein folding and interaction with ligands and receptors. N-Glycosylation trees are complex structures of linked carbohydrate residues attached to asparagine residues. While carbohydrates are typically modeled in protein structures, they are often incomplete or have the wrong chemistry. Here, new tools are presented to automatically rebuild existing glycosylation trees, to extend them where possible, and to add new glycosylation trees if they are missing from the model. The method has been incorporated in the PDB-REDO pipeline and has been applied to build or rebuild 16 452 carbohydrate residues in 11 651 glycosylation trees in 4498 structure models, and is also available from the PDB-REDO web server. With better modeling of N-glycosylation, the biological function of this important modification can be better and more easily understood.

Characterization and structure determination of a llama-derived nanobody targeting the J-base binding protein 1.

J-base binding protein 1 (JBP1) contributes to the biosynthesis and maintenance of base J (ß-D-glucosylhydroxymethyluracil), a modification of thymidine confined to some protozoa. Camelid (llama) single-domain antibody fragments (nanobodies) targeting JBP1 were produced for use as crystallization chaperones. Surface plasmon resonance screening identified Nb6 as a strong binder, recognizing JBP1 with a 1:1 stoichiometry and high affinity (Kd = 30 nM). Crystallization trials of JBP1 in complex with Nb6 yielded crystals that diffracted to 1.47 Šresolution. However, the dimensions of the asymmetric unit and molecular replacement with a nanobody structure clearly showed that the crystals of the expected complex with JBP1 were of the nanobody alone. Nb6 crystallizes in space group P31 with two molecules in the asymmetric unit; its crystal structure was refined to a final resolution of 1.64 Å. Ensemble refinement suggests that in the ligand-free state one of the complementarity-determining regions (CDRs) is flexible, while the other two adopt well defined conformations.

Homology-based loop modeling yields more complete crystallographic protein structures.

Inherent protein flexibility, poor or low-resolution diffraction data or poorly defined electron-density maps often inhibit the building of complete structural models during X-ray structure determination. However, recent advances in crystallographic refinement and model building often allow completion of previously missing parts. This paper presents algorithms that identify regions missing in a certain model but present in homologous structures in the Protein Data Bank (PDB), and 'graft' these regions of interest. These new regions are refined and validated in a fully automated procedure. Including these developments in the PDB-REDO pipeline has enabled the building of 24 962 missing loops in the PDB. The models and the automated procedures are publicly available through the PDB-REDO databank and webserver. More complete protein structure models enable a higher quality public archive but also a better understanding of protein function, better comparison between homologous structures and more complete data mining in structural bioinformatics projects.

Making glycoproteins a little bit sweeter with PDB-REDO.

Glycosylation is one of the most common forms of protein post-translational modification, but is also the most complex. Dealing with glycoproteins in structure model building, refinement, validation and PDB deposition is more error-prone than dealing with nonglycosylated proteins owing to limitations of the experimental data and available software tools. Also, experimentalists are typically less experienced in dealing with carbohydrate residues than with amino-acid residues. The results of the reannotation and re-refinement by PDB-REDO of 8114 glycoprotein structure models from the Protein Data Bank are analyzed. The positive aspects of 3620 reannotations and subsequent refinement, as well as the remaining challenges to obtaining consistently high-quality carbohydrate models, are discussed.

Homology-based hydrogen bond information improves crystallographic structures in the PDB.

The Protein Data Bank (PDB) is the global archive for structural information on macromolecules, and a popular resource for researchers, teachers, and students, amassing more than one million unique users each year. Crystallographic structure models in the PDB (more than 100,000 entries) are optimized against the crystal diffraction data and geometrical restraints. This process of crystallographic refinement typically ignored hydrogen bond (H-bond) distances as a source of information. However, H-bond restraints can improve structures at low resolution where diffraction data are limited. To improve low-resolution structure refinement, we present methods for deriving H-bond information either globally from well-refined high-resolution structures from the PDB-REDO databank, or specifically from on-the-fly constructed sets of homologous high-resolution structures. Refinement incorporating HOmology DErived Restraints (HODER), improves geometrical quality and the fit to the diffraction data for many low-resolution structures. To make these improvements readily available to the general public, we applied our new algorithms to all crystallographic structures in the PDB: using massively parallel computing, we constructed a new instance of the PDB-REDO databank (https://pdb-redo.eu). This resource is useful for researchers to gain insight on individual structures, on specific protein families (as we demonstrate with examples), and on general features of protein structure using data mining approaches on a uniformly treated dataset.

Validation and correction of Zn-CysxHisy complexes.

Many crystal structures in the Protein Data Bank contain zinc ions in a geometrically distorted tetrahedral complex with four Cys and/or His ligands. A method is presented to automatically validate and correct these zinc complexes. Analysis of the corrected zinc complexes shows that the average Zn-Cys distances and Cys-Zn-Cys angles are a function of the number of cysteines and histidines involved. The observed trends can be used to develop more context-sensitive targets for model validation and refinement.

Data Mining of Macromolecular Structures.

The use of macromolecular structures is widespread for a variety of applications, from teaching protein structure principles all the way to ligand optimization in drug development. Applying data mining techniques on these experimentally determined structures requires a highly uniform, standardized structural data source. The Protein Data Bank (PDB) has evolved over the years toward becoming the standard resource for macromolecular structures. However, the process selecting the data most suitable for specific applications is still very much based on personal preferences and understanding of the experimental techniques used to obtain these models. In this chapter, we will first explain the challenges with data standardization, annotation, and uniformity in the PDB entries determined by X-ray crystallography. We then discuss the specific effect that crystallographic data quality and model optimization methods have on structural models and how validation tools can be used to make informed choices. We also discuss specific advantages of using the PDB_REDO databank as a resource for structural data. Finally, we will provide guidelines on how to select the most suitable protein structure models for detailed analysis and how to select a set of structure models suitable for data mining.