Analysis of the Quality of NMR Protein Structures

1
Analysis of the Quality of NMR Protein Structures

• With A Structure Calculated From Your NMR Data,
  How Do You Determine the Accuracy and Quality of
  the Structure?
• Consistency with Known Protein Structural
  Parameters
• bond lengths, bond angles, dihedral angles, VDW
  interactions, etc
• all the structural details discussed at length
  in the beginning
• Consistency with the Experimental DATA
• distance constraints, dihedral constraints,
  RDCs, chemical shifts, coupling constants
• all the data used to calculate the structure
• Consistency Between Multiple Structures
  Calculated with the Same Experimental DATA

Overlay of 30 NMR Structures
2
Analysis of the Quality of NMR Protein Structures

• As We have seen before, the Quality of X-ray
  Structures can be monitored by an R-factor
• No comparable function for NMR
• Requires a more exhaustive analysis of NMR
  structures

3
Analysis of the Quality of NMR Protein Structures

• Root-Mean Square Distance (RMSD) Analysis of
  Protein Structures
• A very common approach to asses the quality of
  NMR structures and to determine the relative
  difference between structures is to calculate an
  rmsd
• an rmsd is a measure of the distance separation
  between equivalent atoms
• two identical structures will have an rmsd of 0Å
• the larger the rmsd the more dissimilar the
  structures

0.81 0.09 Å for all atoms
0.43 0.06 Å for the backbone atoms
4
Analysis of the Quality of NMR Protein Structures

• Root-Mean Square Distance (RMSD) Analysis of
  Protein Structures
• A variety of approaches can be used to measure
  an RMSD
• only backbone atoms
• exclude disordered regions
• only regions with defined secondary structure
• only the proteins active-site region
• on a per-atom or per-residue basis

rmsd difference between NMR and X-ray structure
5
Analysis of the Quality of NMR Protein Structures

• Root-Mean Square Distance (RMSD) Analysis of
  Protein Structures
• Using XPLOR to measure an RMSD
• only one PSF file can be used
• the structures being compared have to be
  consistent with the PSF file
• can only compare similar structures/sequence
• can not compare structural homologs

structure _at_PROTEIN.psf end set precision 3
end evaluate (1 PROTEIN_ave.min") set
displayall_all.rms end display 1 set
displayback_all.rms end display 1 coor
dispcomp _at__at_1 for 2 in ( PROTEIN_dg_103.sam"
PROTEIN_dg_3.sam . . . PROTEIN_dg_18.sam" PROTE
IN_dg_84.sam" )
Read in the PSF file and identify the reference
structure
Name the files to contain the rmsd data
Compare the coordinates
Loop through a list of files to compare against
PROTEIN_ave.min
6
Analysis of the Quality of NMR Protein Structures

Continuation of XPLOR Script
The following set of commands are executed for
each protein in the list
loop fil2 coor _at__at_2 coor
select ((not hydro) and (resid
481 or resid 90111)) fit
end coor select
((not hydro) and (resid 481 or resid 90111))
rms end evaluate
(20result) set display all_all.rms
end display 2 20 coor
select ((name ca or name
c or name n) and (resid 481 or resid 90111))
fit end coor
select ((name ca or name c
or name n) and (resid 481 or resid 90111))
rms end set display
back_all.rms end evaluate (20result)
display 2 20 end loop fil2 stop
First, the structures are aligned based on the
selected atoms
Then, an rmsd is calculated based on the selected
atoms and written to the specified file
Selection tool is very flexible and you can
choose a variety of atom names, types, residue
names and types and chains
7
Analysis of the Quality of NMR Protein Structures

• Some Caveats About (RMSD) Analysis of Protein
  Structures
• rmsd is a global measure
• an unexpected high or low value may reflect only
  a region of the protein that is
  ordered/disordered or similar/dissimilar
• For an ensemble of NMR structures, rmsd implies
  precision
• accuracy can be inferred by comparing structures
  determined by multiple methods and groups
• High rmsd is an indicator of protein dynamics
• This is generally true, but it strongly depends
  on the completeness and accuracy of analyzing the
  NOESY data.
• regions of mobility need to be confirmed by
  experimental measurments
• high rmsd is directly related to the number of
  distance constraints

8
Analysis of the Quality of NMR Protein Structures

• Literature Reports of NMR Protein Structures
• RMSD provides a means to evaluate the quality of
  the structures
• Always calculate multiple structures (10-100)
  with a single experimental data set
• identify the lowest energy structures (10-30)
  ltSAgt
• calculate an average structure (SA)
• minimize the average structure (SA)r against
  the NMR experimental constrains
• calculate rmsd between the ensemble of
  structures (ltSAgt), (SA), (SA)r and
• x-ray structure (if available)

9
Analysis of the Quality of NMR Protein Structures

• What is an Average NMR Structure?
• Average the x,y,z-coordinate positions for each
  atom over the entire ensemble of structures

Structure 1 ATOM 1 N MET 1
0.825 20.945 0.300 Structure 2 ATOM 1 N
MET 1 3.524 22.037 -0.733 Structure
3 ATOM 1 N MET 1 -0.813 21.297
-1.463 . . .
Align the structures and calculate an average
Average Structure ATOM 1 N MET 1
2.888 21.188 -0.390
10
Analysis of the Quality of NMR Protein Structures

• Is the Average NMR Structure a Real Structure?
• No-it is a distorted structure
• level of distortions depends on the similarity
  between the structures in the ensemble
• provides a means to measure the variability in
  atom positions between an ensemble of structures

Expanded View of an Average Structure
Some very long, stretched bonds
Position of atoms are so scrambled the graphics
program does not know which atoms to draw bonds
between
Some regions of the structure can appear
relatively normal
11
Analysis of the Quality of NMR Protein Structures

• Refine or Minimize the Average NMR Structure
• Remove bad bond lengths, angles, etc
• Minimize against the experimental NMR data
• Is this structure a true representation of the
  average of the ensemble of structures?
• A point of discussion in the NMR community
• an overlay of the refined average structure with
  the ensemble of structures will place the average
  in the center of the structures
• Alternative view is to select one structure from
  the ensemble
• Does this represent the average of the ensemble?

Overlay of backbone atoms for the refined
minimized structure (blue) and ensemble of 30
structures (yellow)
12
Analysis of the Quality of NMR Protein Structures

• Structural Statistics Provide Another Approach to
  Quantify the Quality of an NMR Structure
• r.m.s deviations from experimental restraints
• energies
• NOE violations, torsional angle violations,
  repulsive forces, Lennard-Jones electrostatics
  (not a target function in refinement)
• deviations from idealized covalent geometry
• PROCHECK
• G-factor, Ramachandran, bad contacts hydrogen
  bond energy

13
Analysis of the Quality of NMR Protein Structures

Energies, violations and rms deviations are
routinely printed as headers in XPLOR Structure
files
REMARK FILENAMEPROTEIN.sam"REMARK

REMARK overall,
bonds, angles, improper,vdw, REMARK energies
-580.981, 51.7637, 218.963, 42.2728, 177.029,
REMARK cdih, noe,
coup, shift, rama,
coll REMARK energies 19.8689, 104.49, 18.5721,
62.6244, -1404.57, 128.003 REMARK

REMARK
bonds, angles, impropers, cdih,
noe, coup REMARK RMS
5.096347E-03, 0.616985, 0.51284, 1.21754,
3.558853E-02, 0.591961 REMARK shifts RMS a, b
0.883305, 0.89601 REMARK
REMARK
cdih coup noe REMARK violations
3 7 3 REMARK shifts 55 REMARK

REMARK REMARK jcoup rms-d
0.591961 REMARK
REMARK

REMARK
REMARK
DATE16-Feb-04 175026 created by user
14
Analysis of the Quality of NMR Protein Structures

Structural Statistics Provide Another Approach to
Quantify the Quality of an NMR Structure
. . . cons inter (all) (all) end print threshold
0.01000 collapse evaluate (rms_colresult) eva
luate (violations_colviolations) print
threshold 0.10000 noe evaluate
(rms_noeresult) evaluate (violations_noeviol
ations) print threshold 1.0000 cdih evaluate
(rms_cdihresult) evaluate (violations_cdihvi
olations) print threshold 0.0500 bonds evaluate
(rms_bondsresult) print threshold 10.0000
angles evaluate (rms_anglesresult) evaluate
(violations_anglesviolations) print threshold
3.0000 impropers evaluate (rms_impropersresul
t) evaluate (violations_impropviolations) . . .
Sets the thresholds for which violations will be
reported
Obtaining rms deviations of experimental
restraints, energies and deviations from
idealized geometry by simply printing the values
from XPLOR
15
Analysis of the Quality of NMR Protein Structures

. . . spectrum 1 restraint 2745
set-i-atoms 93 ALA
HN set-j-atoms 78 VAL O
Rltaveragegt 2.469 NOE 1.80 (- 0.30/ 0.50)
Delta -0.169 E(NOE) 1.432
spectrum 1 restraint 2747
set-i-atoms 76 GLY HN
set-j-atoms 95 VAL O
Rltaveragegt 2.457 NOE 1.80 (- 0.30/ 0.50)
Delta -0.157 E(NOE) 1.230 NOEPRI RMS
diff. 0.040, (violat.gt 0.1) 69 of
2750 NOEs NOEPRI RMS diff. class INTR
0.004, (viol.gt 0.1) 0 of 483 NOEs
NOEPRI RMS diff. class SHOR 0.035, (viol.gt
0.1) 22 of 825 NOEs NOEPRI RMS diff.
class LONG 0.040, (viol.gt 0.1) 20 of
651 NOEs NOEPRI RMS diff. class SEQ 0.046,
(viol.gt 0.1) 10 of 719 NOEs NOEPRI RMS
diff. class H-BO 0.104, (viol.gt 0.1) 17
of 72 NOEs X-PLORgtevaluate
(rms_noeresult) EVALUATE symbol RMS_NOE
set to 0.399183E-01 (real) X-PLORgtevaluate
(violations_noeviolations) EVALUATE symbol
VIOLATIONS_NOE set to 69.0000 (real). . .
Typical Output of an XPLOR restraint analysis
where violated restraints are listed followed by
a summary
Violated NOE restraint, average distance, error
and energy
RMS summary and number of violated restraints
16
Analysis of the Quality of NMR Protein Structures

• As We Discussed Before, PROCHECK is a Very
  Valuable Tool For Accessing The Quality of a
  Protein Structure
• Correct f, y, c1, c2 distribution
• Comparison of main chain and side-chain
  parameters to standard values

17
Analysis of the Quality of NMR Protein Structures

• Predict NOEs from the Experimental Structure and
  Compare to the NOESY Data
• short distances observed in the structure should
  correlate with NOE cross-peaks
• missing NOE peaks may arise because of
• degeneracy
• too close to diagonal
• overlap with noise
• disordered region of the structure

set-i-atoms 2 THR HN
set-j-atoms 1 MET HT1
1 MET HT2 1 MET
HT3 ( ( ltR-6gt )-1/6 )
1.684
set-i-atoms 2 THR HN
set-j-atoms 1 MET HA ( (
ltR-6gt )-1/6 ) 3.404

set-i-atoms 2 THR HN
set-j-atoms 1 MET HB1
1 MET HB2 ( ( ltR-6gt )-1/6 )
4.137

while (3 lt 113) loop select evaluate (4 3
1) noe set display 2 end set print 2 end
predict from(resid 3 and name h)
to(resid 4113 and name h ) cutoff4.5
end end evaluate (331) end
loop select end loop dist
Lists all the predicted NOEs between the selected
atom pairs within the given distance cut-off
18
Analysis of the Quality of NMR Protein Structures

• NMR R-factor
• difference between expected and observed NOEs
• expected NOEs ? structure
• observed NOEs ? NMR spectra
• also includes unassigned NOEs
• perfect fit would yield R 0
• R-factors have not been readily adapted in NMR
  community
• affected by completeness of assignments, peak
  overlap, sensitivity, noise, extent of data
  (RDCs, coupling constants, etc
• trends with rmsd without complications

Journal of Biomolecular NMR, 17 137151, 2000.