• •

• • 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