Structural databases
 Structural databases
 3D data validation
 3D protein modelling
 Models validation and databases
Outline
Structural databases & Models of structures 2
 Structural databases
 Data formats (PDB, mmCIF, PDBML)
 wwPDB
 Other resources
 3D data validation
 3D protein modelling
 Models validation and databases
Outline
34-Str. DBs & 3D Modelling -> Str. DBs
Data formats
4-Str. DBs & 3D Modelling -> Str. DBs
 different formats are used to represent primary
macromolecular 3D structure data
 PDB
 mmCIF
 PDBML
 ...
 The spatial 3D coordinates for each atom are recorded
4
PDB format
 designed in the early 1970s - first entries of PDB database
 rigid structure of 80 characters per line, including spaces
 still the most widely supported format
4-Str. DBs & 3D Modelling -> Str. DBs -> PDB format 5
PDB format
4-Str. DBs & 3D Modelling -> Str. DBs -> PDB format 6
PDB format
 atomic coordinates
 chemical and biological features
 experimental details of the structure determination
 structural features
 secondary structure assignments
 hydrogen bonding
 biological assemblies REMARK 350
 active sites
 ...
4-Str. DBs & 3D Modelling -> Str. DBs -> PDB format 7
PDB format
 advantages
 widely used → supported by majority of tools
 easy to read and easy to use
→ suitable for accessing individual entries
4-Str. DBs & 3D Modelling -> Str. DBs -> PDB format 8
PDB format
 disadvantages
 inconsistency between individual PDB entries as well as PDB
records within one entry (e.g., different residue numbering in
SEQRES and ATOM sections) → not suitable for computer extraction
of information
4-Str. DBs & 3D Modelling -> Str. DBs -> PDB format 9
 disadvantages
 inconsistency between individual PDB entries as well as PDB
records within one entry → not suitable for computer extraction of
information
 absolute limits on the size of certain items of data, e.g.: max.
number of atom records limited to 99,999; max. number of chains
limited to 26 → large systems such as the ribosomal subunit must be
divided into multiple PDB files
→ not suitable for analysis and comparison of experimental
and structure data across the entire database
PDB format
4-Str. DBs & 3D Modelling -> Str. DBs -> PDB format 10
mmCIF format
 macromolecular Crystallographic Information File (mmCIF)
 developed to handle increasingly complicated structure data
 each field of information is explicitly assigned by a tag and
linked to other fields through a special syntax
4-Str. DBs & 3D Modelling -> Str. DBs -> mmCIF format 11
mmCIF format
 advantages
 easily parsable by computer software
 consistency of data across the database
 disadvantages
 difficult to read
 rarely supported by visualization and computational tools
→ suitable for analysis and comparison of experimental and
structure data across the entire database
→ not suitable for accessing individual entries
4-Str. DBs & 3D Modelling -> Str. DBs -> mmCIF format 12
PDBML format
 Protein Data Bank Markup Language (PDBML)
 XML version of PDB format
4-Str. DBs & 3D Modelling -> Str. DBs -> PDBML format 13
Structural databases
 Primary
• wwPDB: 3D structure of biopolymers
• BMRB: Nuclear Magnetic Resonance specific
• EMDB: Electron-Microscopy specific
• NDB: 3D structure of nucleic acids: http://ndbserver.rutgers.edu/
• CSD: 3D structure of small molecules (commercial)
http://www.ccdc.cam.ac.uk/products/csd/
 Other sources
• PDBsum, SCOP, Protopedia, Structural Biology KnowledgeBase
4-Str. DBs & 3D Modelling -> Str. DBs 14
wwPDB
 joint initiative of four organizations
 Research Collaboratory for Structural Bioinformatics (RCSB PDB)
 Protein Data Bank in Europe (PDBe)
 Protein Data Bank Japan (PDBj)
 Biological Magnetic Resonance Data Bank (BMRB)
4-Str. DBs & 3D Modelling -> Str. DBs -> wwPDB 15
wwPDB
 database growth
4-Str. DBs & 3D Modelling -> Str. DBs -> wwPDB 16
wwPDB
 worldwide Protein Data Bank (wwPDB)
 http://www.wwpdb.org/
 central repository of experimental macromolecular structures
 more than 225,000 structures (October 2024), updated every week
 mostly protein structures (87 %), structures of protein/nucleic acids or
oligosaccharides complexes (11 %) and nucleic acid structures (2 %)
 majority of structures from X-ray crystallography (84 % ), NMR (6 %), or
EM (10%)
 deposition of the structure into wwPDB is a requirement for its
publication
4-Str. DBs & 3D Modelling -> Str. DBs -> wwPDB 17
wwPDB – data deposition
 All data can be deposited at RCSBPDB, PDBe or PDBj site
 Same requirements content and format of the final files:
 structures of biopolymers
 structures determined by experimental techniques
 structures containing required information
 Same validation methods
→ uniformity of the final archive
 PDB-ID
 assigned to each deposition
 unique identifier of each structure
 four-character code
4-Str. DBs & 3D Modelling -> Str. DBs -> wwPDB 18
wwPDB – data validation
 assessment of the quality of deposited atomic models
(structure validation) and how well these models fit
experimental data (experimental validation)
 validation using accepted community standards
 covalent bond distances and angles
 stereochemical validation
 atom and ligand nomenclature
 geometry
 NMR data specific checks
 ...
4-Str. DBs & 3D Modelling -> Str. DBs -> wwPDB 19
wwPDB – data access
 the access to the PDB archive is free and publicly available
from the RCSB PDB site, PDBe site or PDBj site
 FTP
 RCSB PDB, PDBe and PDBj sites distribute the same PDB archive
 updated weekly
 web sites
 each wwPDB site provides its own services and resources →
different views and analyses of the structural data
 sequence-based and text-based queries
4-Str. DBs & 3D Modelling -> Str. DBs -> wwPDB 20
RCSB PDB
 http://pdb.rcsb.org
4-Str. DBs & 3D Modelling -> Str. DBs -> wwPDB 21
Other structure-based resources
 PDBsum
 http://www.ebi.ac.uk/pdbsum/
 provides summaries and pre-computed analyses for structures
deposited in the wwPDB
4-Str. DBs & 3D Modelling -> Str. DBs -> Other resources 24
Other structure-based resources
 Structural Classification of Proteins (SCOP)
 http://scop.mrc-lmb.cam.ac.uk/scop/
 provides classifications of proteins with known 3D structure
according to their evolutionary and structural relationships
4-Str. DBs & 3D Modelling -> Str. DBs -> Other resources 25
Other structure-based resources
 Proteopedia
 http://www.proteopedia.org/wiki/index.php/
 free, collaborative 3D-encyclopedia of proteins and other molecules
4-Str. DBs & 3D Modelling -> Str. DBs -> Other resources 26
Other structure-based resources
 Structural Biology Knowledgebase
 http://sbkb.org/
 provides up-to-date information about advances in structural
biology and structural genomics
4-Str. DBs & 3D Modelling -> Str. DBs -> Other resources 27
Structural quality assurance
284-Str. DBs & 3D Modelling -> 3D data validation
 Revision of concepts
 Important truths about structures
 Errors in deposited structures
 systematic errors
 random errors
 Selecting reliable structure
 rules of thumbs
 quality checks
 programs and databases
Outline
294-Str. DBs & 3D Modelling -> 3D data validation
Concepts
 Resolution
 measure of the level of detail present in the diffraction pattern
3 Å 2 Å 1 Å
304-Str. DBs & 3D Modelling -> 3D data validation -> Concepts
Concepts
 R-factor (R-value)
 measure of a model quality - i.e. how well it can reproduce
experimental data
314-Str. DBs & 3D Modelling -> 3D data validation -> Concepts
Concepts
 Thermal factors (B-factors)
 measure of how much an atom oscillates or vibrates around the
position specified in the model
324-Str. DBs & 3D Modelling -> 3D data validation -> Concepts
Important truths about structures
 all structures are just models devised to satisfy experimental
data → random and systematic errors
 individual structures differ in the quality
 most structures are reasonably accurate, containing “only”
random errors, but some structures are seriously incorrect
 structures should be carefully selected and critically assessed
before being used for a specific purpose → quality checks of
structures
334-Str. DBs & 3D Modelling -> 3D data validation -> Truths
Errors in deposited structures
 systematic errors
 random errors
344-Str. DBs & 3D Modelling -> 3D data validation -> Errors
Systematic errors
 relate to the accuracy of the model—how well it corresponds
to the “true” structure of the molecule in question
 often include errors of interpretation
 low quality of electron density map → difficult to find the correct
tracing of the molecule(s) through it → misstracing and “frame-shift”
errors
 spectral interpretations (assignment of individual NMR signals to
individual atoms)
 may lead to completely wrong final structure
354-Str. DBs & 3D Modelling -> 3D data validation -> Errors -> Systematic
Examples of systematic errors
 completely wrong structures
 trace of the protein chain following the wrong path through the
electron density → completely incorrect fold
Incorrect model (1PHY) Corrected model (2PHY)
364-Str. DBs & 3D Modelling -> 3D data validation -> Errors -> Systematic
Examples of systematic errors
 wrong connectivity between secondary structure elements
 incorrect order of secondary structure elements → many protein’s
residues in the wrong place in the 3D structure
Incorrect model (1PTE) Corrected model (3PTE)
374-Str. DBs & 3D Modelling -> 3D data validation -> Errors -> Systematic
Examples of systematic errors
 frame-shift errors
 occur where a residue is fitted into the electron density that belongs
to the next residue and persists until compensating error is made
(two residues are fitted into the density of a single residue)
 occur almost exclusively at very low resolution (> 3.0 Å), often in
loop regions
 fitting of incorrect main chain or side chain conformations
into the density
 usually the least serious, however still can have effects on biological
interpretations
384-Str. DBs & 3D Modelling -> 3D data validation -> Errors -> Systematic
Random errors
 depend on how precisely a given measurement can be made
 all measurements contain errors at some degree of precision
→ uncertainties in atomic positions
 less serious than systematic errors
 if a structure is essentially correct, the sizes of the random
errors determine how precise the structure is
394-Str. DBs & 3D Modelling -> 3D data validation -> Errors -> Random
Examples of random errors
 uncertainties in atomic positions
 typically in range of 0.01 - 1.27 Å, median 0.28 Å
404-Str. DBs & 3D Modelling -> 3D data validation -> Errors -> Random
Examples of random errors
 side chain flips
 His/Asn/Gln – symmetrical in terms of shape → fit electron density
equally well when rotated by 180°
difficult to distinguish N and O atoms of the side-chain amide from X-ray data
N O O N
414-Str. DBs & 3D Modelling -> 3D data validation -> Errors -> Random
Selecting reliable structure
 rules of thumb for selecting structures
 X-ray structures
 NMR structures
 quality checks of structures
 validation of protein structures
 programs for quality checks
 quality information on the web
424-Str. DBs & 3D Modelling -> 3D data validation -> Structure Selection
Rules of thumb for selecting structures
 X-ray structures
 reasonably accurate structure: resolution ≤ 2.0 Å and R-factor ≤ 0.2
 selection criteria always depend on the type of analysis required
(e.g., comparison of folds – 3.0 Å resolution is sufficient vs. analysis
of side chain torsional conformers – resolution ≤ 1.2 Å is required)
 R-factor can easily be fooled → a better indicator of model
reliability is Rfree – calculated in the same way as R-factor but using
only a small fraction of the experimental data; Rfree should be ≤ 0.4
 local errors indicated by residue B-factors > 50 but quality checks
should always be performed to assess possible local problems in a
structure
434-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Rules of thumb
Rules of thumb for selecting structures
 NMR structures
 no simple rule of thumb as in the case of X-ray structures
 information on structure quality can be found in the original paper
or obtained by quality checks
 ResProx (http://www.resprox.ca/) – predicts the atomic resolution
of NMR protein structures using machine learning
 DRESS (http://www.cmbi.ru.nl/dress/ ) and RECOORD
(http://www.ebi.ac.uk/pdbe-apps/nmr/recoord/main.html)web
servers – provide improved versions of old NMR models (obtained
by re-refinement of the original experimental data using more upto-date
force fields and refinement protocols)
444-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Rules of thumb
Quality checks of structures
 checks of structure geometry, stereochemistry and other
structural properties
 tests of normality
 comparison of a given protein or nucleic acid structure against what
is already known about these molecules
 knowledge comes from high-resolution structures of small molecules
and systematic analyses of existing protein and nucleic acid
structures
 not all outliers from the norm are errors (e.g., an unusual torsion
angle of a single residue), however, a structure exhibiting a large
number of outliers and oddities is probably problematic
454-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Validation of protein structures
 Ramachandran plot
 check of stereochemical quality of protein structures
 plot of the Ψ versus the Φ main chain torsion angles for every amino
acid residue in the protein (except the two terminal residues)
 favorable and “disallowed” regions of the plot determined from
analyses of existing structures
 typical protein structures – residues tightly clustered in the most
favored regions, only few or none residues in the “disallowed” regions
 poorly defined protein structures– residues more dispersed and many
of them lie in the “disallowed” regions of the Ramachandran plot
464-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Validation of protein structures
 Ramachandran plot
typical protein structure poorly defined protein structure
Φ
Ψ
Φ
Ψ
474-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Validation of protein structures
 Ramachandran plot
484-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
 side chain torsion angles
 preferred conformations of side chain torsion angles obtained by
analyses of existing structures
 χ1 – torsion angle about N-Cα-Cβ-Aγ
 χ2 – torsion angle about Cα-Cβ-Aγ-Aδ, ...
Validation of protein structures
494-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Validation of protein structures
 bad and unfavorable atom-atom contacts
 “simple” count of bad contacts, e.g., two nonbonded atoms with a
center-to-center distance < sum of their van der Waals radii
 evaluation of the environment of individual atoms or residue
fragments with respect to the environments found in the high
resolution crystal structures
504-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Validation of protein structures
 secondary structure
 ~ 50-60% of residues usually in regions of regular secondary structure
 poorly defined structures – main chain O and N atoms can lie beyond
normal hydrogen bonding distances → some of the α-helices and βstrands
not detected by the secondary structure assignment programs
typical protein structure poorly defined protein structure
514-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Validation of protein structures
 other parameters
 counts of unsatisfied hydrogen bond donors
 hydrogen bonding energies
 knowledge-based potentials assessing how “happy” each residue
is in its local environment – many unhappy residues → “sad”
overall structure
 real space R-factor expressing how well each residue fits its
electron density; can also be expressed as a Real-space correlation
coefficient
524-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Programs for quality checks
 Proteins
 PROCHECK
 WHAT_CHECK
 Verify 3D
 MolProbity
 ANOLEA
534-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Programs for quality checks
 PROCHECK
 http://www.ebi.ac.uk/thornton-srv/software/PROCHECK/
 variety of plots for protein structures: Ramachandran plot, χ1-χ2
plot for each amino acid type, main chain bond lengths and bond
angles, secondary structure plot, ...
 parameters that deviate from norm are highlighted
 NMR-PROCHECK – version specific for NMR
544-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Programs for quality checks
 WHAT_CHECK (subset of WHAT IF package)
 http://swift.cmbi.ru.nl/gv/whatcheck/
 space group and symmetry
 bond lengths and angles
 bad contacts
 hydrogen bonds
 ....
 detailed output of discrepancies of the given protein structure
from the norms
554-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Programs for quality checks
 Verify3D
 https://genesilico.pl/toolkit/unimod?method=Verify3D
 evaluates residue’s environment in terms of secondary structure,
buried surface area, and fraction of side chain covered by polar atoms
 MolProbity
 http://molprobity.biochem.duke.edu/
 detailed all-atom contact analysis within a given protein structure
 ANOLEA
 http://melolab.org/anolea/index.html
 knowledge based evaluation of atom-atom contacts
564-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Quality information on the web
 several databases provide pre-computed quality criteria for
all wwPDB structures
 EDS
 PDBsum
 PDBREPORT
 RCSB PDB
574-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Quality information on the web
 Electron Density Server (EDS)
 http://eds.bmc.uu.se/eds/, also available via the PDBe site
 information about local quality of the structure for all structures
from wwPDB with deposited experimental data
 plot of real-space R-factor (RSR) – how well each residue fits its
electron density
 plot of Z-score – large positive spike → residue has considerably
worse RSR than the average residue of the same type in structures
determined at similar resolution.
 Ramachandran plot
 ...
584-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Quality information on the web
 Electron Density Server (EDS)
594-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Quality information on the web
 PDBsum
 http://www.ebi.ac.uk/pdbsum/
 provides numerous structural analyses of all wwPDB structures,
including full PROCHECK output (for all protein-containing entries)
604-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Quality information on the web
 PDBsum
614-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Quality information on the web
 PDBREPORT
 http://swift.cmbi.ru.nl/gv/pdbreport/
 provides a pre-computed WHAT_CHECK report for any structure in
the wwPDB
624-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Quality information on the web
 PDBREPORT
634-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
Quality information on the web
 RCSB PDB
 http://pdb.rcsb.org/
 provides geometrical analyses for each entry, including information
about bond lengths, angles and dihedral angles
644-Str. DBs & 3D Modelling -> 3D data validation -> Str. Sel. -> Quality checks
References
 Gu, J. & Bourne, P. E. (2009). Structural Bioinformatics, 2nd Edition,
Wiley-Blackwell, Hoboken, p. 1067.
 Xiong, J. (2006). Essential Bioinformatics. Cambridge University Press,
New York, p. 352.
 Schwede, T. & Peitsch, M. C. (2008). Computational Structural Biology:
Methods and Applications, World Scientific Publishing Company,
Singapore, p. 700.
 Shapiro, B. A. et al. (2007). Bridging the gap in RNA structure prediction.
Current opinion in structural biology 17: 157-165.
65Structural databases & Models of structures