Validation of NMRderived structures

1
Validation of NMR-derived structures

• Chris Spronk
• Centre for Molecular and Biomolecular Informatics
• University of Nijmegen
• The Netherlands

2
Overview

• Introduction
• Selection of structures
• Validation of structures using experimentally
  derived restraints
• Validation of structures using local and overall
  structure quality indicators

3
Why validate structures?

• Bio-macromolecular structures are a valuable
  source for understanding biology
• Structure based drug design
• Homology modeling
• Structural genomics
• Structures should be reliable
• Satisfy experimental data
• Good local and overall quality

4
NMR Structure determination
NMR experimental data
Structure ensemble
Experimental restraints
Structure calculation and selection
Assignment and conversion
Restraint violation and error analysis
Validated structure data
Structure quality checks and statistics
(often not done!)
5
Selection of NMR structures (1)

• Common selection criteria
• Violation cutoff criteria, e.g.
• No distance restraint violations gt 0.5Å
• No dihedral angle violations gt 5
• Energy criteria e.g.
• Select a sub-ensemble consisting of the lowest
  energy structures

6
Selection of NMR structures (2)
rmsd3.04
rmsd0.82
rmsd0.77
7
Satisfaction of experimental data

• Number and size of violations
• Fit of the structure to the experimental data
• Root mean square deviations
• NOE, Dihedral angles etc.
• Cornilescu Q-factors
• Residual dipolar couplings
• Experimental restraint energy

8
Evaluation of experimental data (1)

• Number of restraints
• Completeness

9
Evaluation of experimental data (2)

• Restraints per residue
• Completeness
• per residue

10
Evaluation of experimental data (3)

• Redundancy of
• restraints

11
Evaluation of experimental data (4)

• Intra-residual
• Side chain conformation
• Sequential residue i to residue i1
• Secondary structure
• Medium range residue i to residue i5
• Secondary structure
• Long range residue i to residue gt i5
• Secondary and tertiary structure

12
Evaluation of experimental data (5)

• New method for evaluation of distance restraints
• QUantitative Evaluation of Experimental NMR
  restraints QUEEN

13
QUEEN (1)

• Quantification of the information contained in
  distance restraints
• Relative contributions of each restraint to the
  determination of a structure
• The method allows to identify
• Important restraints
• Unique restraints (?error analysis)
• Redundant restraints

14
QUEEN (2)
YBOX 1h95
10 most important restraints
15
QUEEN (3)
1GB1 Rather unique
3GB1 Less unique
IgG binding domain (1gb1)
X-ray structure
Restraint Ala26 Tyr45
16
QUEEN (4)
17
R-factors and cross-validation

• R-factor
• Free R-factor (X-ray)
• Calculate structures with 90 of data (working
  set)
• Determine free R on 10 of data (test set)
• Complete cross validation (NMR)
• Use a number of different randomly chosen test
  sets

18
Uncertainty in structure coordinates

• X-ray crystallography B-factor
• Quality of the crystal
• Dynamic behavior of the molecule
• Disorder
• NMR atomic Root Mean Square Deviation or RMSD
• Should reflect measured dynamics and the
  uncertainty in the experimental data
• Used as a measure for precision and accuracy

19
Coordinate RMSDs

• Calculation requires superposition of structures
• Region dependent
• Use Circular Variance, CV?
• Structure selection criteria are subjective
• Allowed variation in structures depends on the
  force-field

20
RMSD and Circular Variance
Arg55
Ser74
21
Precision versus Accuracy (1)

• Precision is the variation of X around ltXgt
• expressed as standard deviation or variance
• Accuracy is the closeness of ltXgt to the true
  value of X
• Precision and accuracy are often mixed in the
  literature

22
Precision versus Accuracy (2)
Precise, not accurate
Accurate, not precise
Precise and accurate
Not accurate and not precise
23
Accuracy of NMR structures

• Accuracy can only be assessed when the true
  structure is known (Gold Standard)
• Only the case for simulated data-sets
• Sometimes X-ray structures are used
• Different experimental conditions
• Crystal contacts
• In some cases X-ray structures fit NMR data
  better than NMR structures

24
Precision and true variance
Precision underestimates true variance
Precision equals true variance
Precision overestimates true variance
25
Re-sampling of ensembles
Original ensemble (60 structures)
Re-sampled ensemble (60 structures)
Narrow bundle low RMSD (0.38) high
precision Unrealistic error estimate
Wide bundle high RMSD (0.94) low
precision More realistic error estimate
26
Part II
27
Validation of protein structure quality

• What type of properties are important?
• How can we check these properties?
• PROCHECK
• WHAT IF (FULLCHECK)
• How is the quality of the properties expressed
• Z-scores, RMS Z-scores (WHAT IF)

28
Validation criteria for protein structures

• Local geometry
• Bond lengths, bond angles, chirality, omega
  angles, side chain planarity
• Overall quality
• Ramachandran plot, rotameric states, packing
  quality, backbone conformation
• Others
• Inter-atomic bumps, buried hydrogen-bonds,
  electrostatics

29
Bonded geometry
Distorted C?-chirality
L-amino acid
D-amino acid
30
Rotameric states
Eclipsed
Staggered
31
Inter-atomic bumps
Overlap of two backbone atoms
32
Omega angles
Trans-conformation (omega180)
Cis-conformation (omega0)
33
Side chain planarity
Planar ARG side-chain (Good)
34
Internal hydrogen bonding
Internal hydrogen bonding in Crambin
35
Electrostatics
After energy minimization including electrostatics
Bad electrostatics
36
Packing quality
Good packing
Bad packing
37
Backbone Conformation
Very normal
Very unique
38
Ramachandran Plot
Phi and Psi angles
Ramachandran plot
39
PROCHECK WHAT IF

• PROCHECK and PROCHECK_NMR
• Very useful graphical and text output
• WHAT IF
• More checks and more critical checks
• The reference data base of X-ray structures is
  continuously updated

40
The WHAT IF reference set

• Overall quality
• Well refined high resolution X-ray structures
  (resolution lt 2.0 Ångstrom, R-factor lt 19)
• Continuously updated
• Local geometric quality
• Cambridge small molecule database (CSD)
• Well refined high resolution X-ray structures

41
A WHAT IF summary report
42
Normal distributions and Z-scores
43
Normal distributions andRMS Z-scores
RMS Z-score0.5
RMS Z-score1.0 (reference)
RMS Z-score2
44
Z-scores and RMS Z-scores

• Structure Z-scores
• Z-scores gt 0 are better than average
• Z-scores lt 0 are worse than average
• However A Z-score of -1 is equally likely as a
  Z-score of 1!!
• Local geometry RMS Z-scores
• Too tight restraining of geometry ? RMS Z-score lt
  1
• Too loose restraining of geometry ? RMS Z-score gt
  1
• Proper Gaussian distribution ? RMS Z-score 1

45
Z-scores and dihedral angle distributions
Ramachandran
Chi-1/Chi-2
Z-score 1.8
Z-score 1.9
Z-score -8.5
Z-score -5.3
46
RMS Z-scores and bond and angle distributions
RMS Z-score 0.96
RMS Z-score 0.97
RMS Z-score 0.22
RMS Z-score 0.01
47
X-ray versus NMR structures
48
NMR structures at the PDB
49
Improving protein NMR structures

• Structures can be significantly improved by final
  refinement in explicit water
• Currently a data base of refined NMR structures
  is built at the CMBI

50
Data base potentials

• Improves the appearance of the quality of
  structures
• Ramachandran plot refinement
• Chi-1/Chi-2 rotamer refinement
• Use with caution!

51
Practicals

• The practicals are focused on the use of WHAT IF
  and PROCHECK only Use these two program to have
  a look at pdb-entries 1i1s and 1ka3. Want went
  wrong in these structures?

52
Acknowledgements

• Nijmegen University Sander Nabuurs
• Elmar Krieger
• Gert Vriend
• Geerten Vuister
• BioMagResBank Jurgen Doreleijers
• Utrecht University Aart Nederveen
• Alexandre Bonvin
• EBI, Cambridge Wim Vranken