Graphics Strategies in NMRPipe• •23 Matrix Decomposition and Non-Uniform Sampling Maximum...
Transcript of Graphics Strategies in NMRPipe• •23 Matrix Decomposition and Non-Uniform Sampling Maximum...
• •
• • 1
Spectrometer format conversion and 1D-4D ComplexFourier Transform and Signal Enhancement
Spectral Visualization
1D-4D Peak Detection and Quantification: Position,Amplitude, Width and Modulation/Evolution
Resonance Assignment
Extraction of Structural Parameters
Distance (NOE) Assignment
Molecular Structure Calculation
Molecular Display and Structure Verification
Exploitation of Structure
SubstantialFacilities
UsefulContributions
Still to do, or In-progress
Ad Bax _ James Chou _ Gabriel Cornilescu _ Alex Grishaev StephanGrzesiek _ Georg Kontaxis _ John Kuszewski _ John Pfeiffer Tobias
Ulmer _ Gerteen Vuister _ Justin Wu _ Guang Zhu _Yang Shen
Graphics Strategies inNMRPipe
• •
• • 2
Graphics Strategies of Edward Tuftewww.edwardtufte.com
Above all, Show the Data
Show Cause and Effect
Represent Data and Scale Faithfully
Maximize “Data Ink” and Data Density, Minimize “Chart Junk”
Shrink Graphics - Integrate Text, Values, and Graphics - BeMultivariate
Use Layers – Use Macro and Micro Interpretations - Clarify byAdding Detail
Conserve Color Space
Use Small Multiples
Find Ways to Show All of the Data
Treat Design as a Solved Problem, then Find the BestExamples
• •
• • 3
Napolean’s Route: 422,000 Men to 10,000 Men, Five Dimensions
• •
• • 4
• •
• • 5
Bax Group Figure: 18 values
Weather Statistics: 1,800+ Values, Four Variables, Notations
• •
• • 6
• •
• • 7
• •
• • 8
• •
• • 9
• •
• • 10
• •
• • 11
NMR Signal Processing
• •
• • 12
• •
• • 13
• •
• • 14
Matrix Decomposition and LP
• •
• • 15
• •
• • 16
• •
• • 17
• •
• • 18
• •
• • 19
• •
• • 20
Non-Uniform Sampling
• •
• • 21
• •
• • 22
Matrix Decomposition and Non-Uniform Sampling
X X
• •
• • 23
Matrix Decomposition and Non-Uniform Sampling
Maximum Likelihood Frequency Map
F(t) – λ exp( αt ) cos( 2πft)
For a given f, choose λ to minimize RMS.
RMS has minima when f matches afrequency in F(t)
Adjust via zero-order baseline correction.
• •
• • 24
MEM and NUS
• •
• • 25
Reproducibility of Shifts
HN Hz 15N Hz 13C Hz
LP 0.67 0.79 2.28
PCA LP 0.67 0.24 3.18
MEM 0.56 0.55 1.90
FDM 3.23 2.47 7.56
Structural Data from NMR
• •
• • 26
Chemical Shift J-Coupling
NOE Distance
• •
• • 27
• •
• • 28
• •
• • 29
Chemical Shift and Backbone Structure Motif
Match database triplet with target, based on sum-of squaresdifference in chemical shifts, plus residue type homology term.
Use central residue as predictor of phi and psi.
• •
• • 30
• •
• • 31
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
C H S T A C D E F G H I K L M N P Q R S T V W Y c
RM
S(P
red
, Ob
s) [
pp
m]
N
HA
C'
CA
CB
HN
15N: 2.34, HA: 0.26, C’: 0.99, CA: 0.88, CB: 0.97 HN:0.46 ppm
• •
• • 32
Alignment by Liquid Crystal
• •
• • 33
• Search PDB for small fragments whosesimulated dipolar couplings and shifts match theobserved values.
• Use the fragment information to reconstitutelarger structural elements.
• Also: Sequential NOEs, J values, etc.• Nucleic Acid Applications
Molecular Fragment Replacement(MFR)
• •
• • 34
Phi-Psi Trajectory of Homology Fragments
Chemical Shifts Dipolar Couplings Known Structure
• •
• • 35
Initial Structure fromAverage Phi and Psi ofFragment Ensemble
1ubq vs MFR phi/psirefined structure
MFR Estimation of Tensor Parameters
• Magnitude
• Rhombicity
• Orientation (Euler Angles)
• •
• • 36
Gamma S
• 177 Residues, two similar domains,homologous structure is known.
• 179 Amide-Amide NOEs, 70 Methyl-Methyl NOEs, including 6 inter-domain
• DC Medium 1: 144 HN-N, 111 CA-CB,150 CA-C’, 130 N-C’
• DC Medium 2: 147 HN-N, 135 CA-CB,153 CA-C’,
• •
• • 37
• Conduct MFR Search with SVD (free tensor)
• Conduct second MFR Search with fixedtensor Da, Rh, and relative orientation
• Refine all fragments with fixed tensor Da, Rh
• Phi and Psi for 90% of residues; 50% havebetter than 5 degree RMS consenus; 33%are 3 degree RMS or better.
dynReadGMC -gmc $gmcDir -pdb $pdbName
for {set i 1} {$i <= $count} {incr i} \ { dynSimulateAnnealing -graph -print 50 -rasmol 500 \ -sa stepCount init 100 \ stepCount high 24000 \ stepCount cool 8000 \ timeStep all 3 \ temperature all 4000 \ temperature coolEnd 0 \ -fc dc coolEnd 2.0 \ torsion all 50 \ torsion coolEnd 10 \ noe all 25 \ noe coolEnd 100 \ radGyr all 0.0
set outName [format $outTemplate $i]
dynWrite -pdb -src $dynInfo(gmc,pdb) -out $outName –rem $dynInfo(energyText) dynRead -pdb -src $dynInfo(gmc,pdb) -in $pdbName
incr iseed 111 srand $iseed }
• •
• • 38
• •
• • 39
NMR Applications in DrugDiscovery
• •
• • 40
NMR Spectral Series:Two Approaches
Applications of NMR in theDrug Discovery Process
SAR by NMR (Abbott Labs)
Observe Ligand SignalsObserve Protein Signals
• •
• • 41
Amide-HN Chemical Shift of Residue i Am
ide-
N C
hem
ical
Shi
ft of
Res
idue
i
• •
• • 42
Analyze Titration Curve to Estimate Kd
Entire spectrum is a singleobject in multdimensionalspace.
Coordinates of the object arethe spectral intensities.
Similar spectra clustertogether.
Spectra with similar featureslie along lines and curves
• •
• • 43
Screen Many Samples or Mixtures for Binding(Display: PCA Method of Roche)
• •
• • 44
• •
• • 45
Identification of Stereochemistryby Dipolar Couplings
RMSD
ISOMER
-1.0
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
ABCD
RMSD
ISOMER
-1.0
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
ABCD
J.L. Yan, F. Delaglio, A. Kaerner, A.D. Kline, H.P. Mo, M.J. Shapiro, T.A. Smitka, G.A. Stephenson, E.R.Zartler: Complete relative stereochemistry of multiple stereocenters using only residual dipolar
couplings. J. Am. Chem. Soc., 126 (15) 5008-5017 (2004).
• •
• • 46
• •
• • 47
Resonance Assignment of Known Structures:
Permutation of PCS and DC Values