fundamental of automated structure determination · 2016. 5. 19. · sp :auto-rickshaw • what...
TRANSCRIPT
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Santosh Panjikar
Fundamental ofAutomated Structure Determination
SAD
MR
MRSIRAS
SIRASMRMAD
MAD
MRRIP
RIP
S-SADMRSAD
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
From Crystal to Structure
Data processing+scaling
HKL suitesMosflm/Scala
Automar/marscaleXDS
D*trek
Derivatisation SeMet
Halide soakingHeavy atom soaking
Xenon/Krypton
Cryo-protection Organic solvent
Oil based Cryo-salt
Structure Solution
DiffractionData collection strategy BEST
How many degrees ?How much dose ?What resolution ?
What detector distance ?………?
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
From Crystal to Structure
Data processing+scaling
HKL suitesMosflm/Scala
Automar/marscaleXDS
D*trek
Derivatisation? SeMet
Halide soakingHeavy atom soaking
Xenon/Krypton
Cryo-protection Organic solvent
Oil based Cryo-salt
Structure Solution
DiffractionData collection strategyBEST
How many degrees ?How much dose ?What resolution ?
What detector distance ?………?
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
From Crystal to Structure
Data processing+scaling
HKL suitesMosflm/Scala
Automar/marscaleXDS
D*trek
DerivatisationSeMet
Halide soakingHeavy atom soaking
Xenon/Krypton
Cryo-protection Organic solvent
Oil based Cryo-salt
Structure Solution
Diffraction ?Data collection strategy BEST
How many degrees ?How much dose ?What resolution ?
What detector distance ?………?
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
From Crystal to Structure
Data processing+Scaling ? HKL suites
Mosflm/Scala Automar/marscale
XDSD*trek
DerivatisationSeMet
Halide soakingHeavy atom soaking
Xenon/Krypton
Cryo-protection Organic solvent
Oil based Cryo-salt
Structure Solution
Diffraction Data collection strategy BEST
How many degrees ?How much dose ?What resolution ?
What detector distance ?………?
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
From Crystal to Structure
Data processing+Scaling
HKL suitesMosflm/Scala
Automar/marscaleXDS
D*trek
DerivatisationSeMet
Halide soakingHeavy atom soaking
Xenon/Krypton
Cryo-protection Organic solvent
Oil based Cryo-salt
Structure Solution ?
Diffraction Data collection strategyBEST
How many degrees ?How much dose ?What resolution ?
What detector distance ?………?
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
From Crystal to Structure
Data processing+Scaling
HKL suitesMosflm/Scala
Automar/marscaleXDS
D*trek
DerivatisationSeMet
Halide soakingHeavy atom soaking
Xenon/Krypton
Cryo-protection Organic solvent
Oil based Cryo-salt
Structure Solution ?
Diffraction Data collection strategyBEST
How many degrees ?How much dose ?What resolution ?
What detector distance ?………?
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Phasing in Macromolecular Crystallography
How do we get from spots on a screen to a pretty picture of our protein?
BY CALCULATING AMPLITUDES AND PHASES!!
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Phase Determination Methods
1. SIR, SIRAS, MIR, MIRAS(single/multiple isomorphous replacement with anomalous scattering)
2. MAD(multiple wavelength anomalous diffraction)
3. SAD (SAS)(single wavelength anomalous diffraction/scattering)
4. RIP, RIPAS(radiation damage induced phasing with anomalous scattering)
5. MR(molecular replacement)
6. Direct Methods
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Data reduction
Location of Heavy atoms
Heavy atom refinementand/or
Phase calculation
Solvent flattening&
NCS averaging
Model building
Model refinement
Expe
rimen
tal P
hasi
ng
Hea
vy a
tom
der
ivat
ives
CCP4
SHELXDCRUNCH2SOLVECNSHYSSSUPERFLIP
SHARPBP3PHASERMLPHARESOLVESHELXE DM
PARROT, PIRATERESOLVE, SOLOMON
ARP/wARPRESOLVEBUCCANNERESSENSALBEMAIDTEXTAL REFMAC5
CNSSHELXL
Manual building
COOTO
Steps in structure determination (Experimental phasing)
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Data reduction
Location of Heavy atoms
Heavy atom refinementand
Phase calculation
Solvent flattening&
NCS averaging
Model building
Model refinement
Steps in structure determination (Experimental phasing)
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Data reduction
Location of Heavy atoms
Heavy atom refinementand
Phase calculation
Solvent flattening&
NCS averaging
Model building
Model refinement
Steps in structure determination (Experimental phasing)
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Primary aim
To achieve an interpretable electron density map anda partial structure in the minimal time in order toconfirm the success of the experiment at thesynchrotron while the crystal is still at or near thebeam line
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Ultimate aim
To achieve a model, which is correct and ascomplete as possible from a seqence file andintensity data at low and high resolution, andachieve almost refined model (~30% Rfree) forthe data better than 3.2 Å resolution.
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
The Auto-Rickshaw Philosophy
Photographed by Prof. B.C. Wang
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
The Auto-Rickshaw Approach
Mimic What An Experienced Crystallographer would Do
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
The Auto-Rickshaw Approach
Mimic What An Experienced Crystallographer would Do
Try To Be As Fast As Possible ... And (Just) As Good As necessary
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
The Auto-Rickshaw Approach
Try to Mimic What An Experienced Crystallographer would Do
Try To Be As Fast As Possible ... And (Just) As Good As necessary
Minimize User Input
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Softwarepipeline
Crystallographic software
Phasing methods
Decision makingand machine learning
Database ofX-ray anomalous
data
Auto-Rickshaw : An Automated Crystal Structure Determination Pipeline
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Decision making process
Decision making module (DMD): Overall decisiong makingSingle choice of software at each step of structure determinationPredefined choice of parameters
SHELXD, SOLVE, CHRUNCH2
SHELXE, MLPHARE, BP3, PHASER, SHARP
DM, PIRROT, RESOLVE, SOLOMON, PIRATE
ARP/wARP, ALBE, RESOLVE, BUCANEER, SHELXE
Master
Step-representative
Program-representative
CPU usage
CCP4 (data reduction utility)
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Master decision makerStep Representative decision makerProgram Representaive decision maker
Decision making
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Decision making
Master decision makerStep Representative decision makerProgram Representaive decision maker
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
The Auto-Rickshaw System for experimental phasing
Phase improvement and model completion
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Panjikar, S., Parthasarathy, V., Lamzin, V. S., Weiss, M.S. & Tucker, P. A. (2005). Auto-Rickshaw - An automatedcrystal structure determination platform as an efficienttool for the validation of an X-ray diffraction experiment.Acta Cryst. D61, 449-457.
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Date: Fri, 11 Feb 2005 10:47:17 -0600From: "Dauter, Zbigniew" To: [email protected]: decision makersParts/Attachments:
Hi Santosh
We just read the proof of your Auto-Rickshaw paper very nice work!One question.
What exactly means (in Abstract) that you combine computer programs withdecision makers?
Does it mean that the intermediate results are rapidly sent to India(Bangalore or Madras) and there some twenty crystallographers makedecisions and transfer them back by telephone?
CheersZbyszek
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
• What should be the maximum resolution cut-off for substructure solution and phasing ?
• When to terminate the task for substructure solution ?
• How many located heavy atom sites are actually correct ?
• Which hand of the substructure is correct ?
• Which is the best way to calculate phases ?
• How to exploit multiple copies (dimers, etc.) if present
• How to get the model?
• How to improve the model ?
Decision making
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
0
0.5
1
1.5
2
2.5
3
3.5
4
8 6 5 4 3.5 3 2.6 2.4 2.2 2 1.8
Resolution cut off
0
10
20
30
40
50
60
70
80
90
100
8 6 5 4 3.5 3 2.6 2.4 2.2 2 1.8
Ano
mal
ous
CC
DA
NO
/σ(I) SAD
HREM/PEAKHREM/INFLHREM/INFL
MAD
Resolution
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
0
0.5
1
1.5
2
2.5
3
3.5
4
8 6 5 4 3.5 3 2.6 2.4 2.2 2 1.8
Resolution cut off
0
10
20
30
40
50
60
70
80
90
100
8 6 5 4 3.5 3 2.6 2.4 2.2 2 1.8
Ano
mal
ous
CC
DA
NO
/σ(I) SAD
HREM/PEAKHREM/INFLHREM/INFL
MAD
Resolution
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
0
0.5
1
1.5
2
2.5
3
3.5
4
8 6 5 4 3.5 3 2.6 2.4 2.2 2 1.8
Resolution cut off
0
10
20
30
40
50
60
70
80
90
100
8 6 5 4 3.5 3 2.6 2.4 2.2 2 1.8
Ano
mal
ous
CC
DA
NO
/σ(I)
Resolution
SAD
HREM/PEAKHREM/INFLHREM/INFL
MAD
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
• What should be the maximum resolution cut-off for substructure solution and phasing ?
• When to terminate the task for substructure solution ?
• How many located heavy atom sites are actually correct ?
• Which hand of the substructure is correct ?
• Which is the best way to calculate phases ?
• How to exploit multiple copies (dimers, etc.) if present
• How to get the model?
• How to improve the model ?
Decision making
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
When to terminate the SHELXD trials
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96
SHELXD TRIALS
CC
(wea
k)
CCweak : Correlation coefficient between observed and calculated E values for weak reflection as computed in SHELXD
SITCOM
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-RickshawSP :Auto-
Substructure determination using SHELXD in Auto-Rickshaw
Initial resolution cut-off based signal/noise ratio (from the program SHELXC).
CCweak: Auto-Rickshaw usually considers better than 15%
If 15% is not achieved in the first round of SHELXD, then resolution is decreased by 0.2 and increase number of trails by 100. Checks the CCweak in the next round.
If CCweak improves, number of heavy atoms sites is increased in the next round of SHELXD trail.
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Substructure determination using SHELXD in AutoRickshaw
Common site using SITCOM (Dall‘Antonia & Schneider)
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
• What should be the maximum resolution cut-off for substructure solution and phasing ?
• When to terminate the task for substructure solution ?
• How many located heavy atom sites are actually correct ?
• Which hand of the substructure is correct ?
• Which is the best way to calculate phases ?
• How to exploit multiple copies (dimers, etc.) if present
• How to get the model?
• How to improve the model ?
Decision making
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Changing hand of the substructure
System Chiral NonchiralTriclinic P1Monoclinic P2, P21, C2Orthorhombic P222, P2221, P21212, P212121, C222,
C2221, I222, I212121, F222Tetragonal P41P43
P4122P4322P41212P43212
P4, P42, I4, I41P422, P4212, P4222, P42212 I422, I4122
Trigonal P31P32P3112P3212 P3121P3221
P3, R3P312, P321, R32
Hexagonal P61P65 P62P64
P6122P6522 P6222P6422
P6, P63
P622, P6322
Cubic P4132P4332 P23, F23, I23, P213, I213P432, P4232, F432,F4132, I432, I4132
For nonchiral space groups the other hand of the heavy-atom sites is found by the operation (x, y, z) (-x, -y, -z), except for three space groups (I41, I4122 , F4132 and I4132) where there is also a change of origin.
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Changing hand of the substructure
I41 (x,y,z) (x+0.5, y, z)I4122 (x,y,z) (x+0.5, y, z+0.25)F4132 (x,y,z) (x+0.75, y+0.25, z+0.75)I4132 (x,y,z) (x+0.25, y+0.25, z+0.25)
For nonchiral space groups the other hand of the heavy-atom sites is found by the operation (x, y, z) (-x, -y, -z), except for three space groups (I41, I4122 , F4132 and I4132) where there is also a change of origin.
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Changing hand of the substructureFor the chiral space groups the change of hand of the heavy-atom sites with the operation (x, y, z)(-x, -y, -z) is accompanied by a change of space group to the other chiral form.
System ChiralTriclinicMonoclinic OrthorhombicTetragonal P41P43
P4122P4322P41212P43212
Trigonal P31P32P3112P3212 P3121P3221
Hexagonal P61P65 P62P64
P6122P6522 P6222P6422
Cubic P4132P4332
Chiral objectsNonsuperimposable
mirror images
Nonchiral objectssuperimposablemirror images
mirror mirror
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Enantiomorph determination
Level-1: ABS SHELXE(C-value) (Contrast)
Level 2: MLPHARE/SHARP/BP3 DM(FOM) ( FOM, FreeR)
Level 3: MAPMAN CAPRA(Maximum Fragment size) (Cα-atoms)
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
• What should be the maximum resolution cut-off for substructure solution and phasing ?
• When to terminate the task for substructure solution ?
• How many located heavy atom sites are actually correct ?
• Which hand of the substructure is correct ?
• Which is the best way to calculate phases ?
• How to exploit multiple copies (dimers, etc.) if present
• How to get the model?
• How to improve the model ?
Decision making
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Example: which software to choose for Phase Calculation ?
CCweak : Correlation coefficient between observed and calculated E values for weak reflection as computed in SHELXD
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Example: which software to choose for Phase Calculation ?
SHELXE
MLPHARE
BP3/SHARP
CCweak : Correlation coefficient between observed and calculated E values for weak reflection as computed in SHELXD
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Determination of correct space groupSystem Space groupMonoclinic P2, P21,
Orthorhombic P222, P2221, P2212, P2122, P21212, P22121, P21221 P212121C222, C2221, I222, I212121,
Tetragonal P4, P42, (P41P43)I4, I41P422, P4212, P4222, P42212 , (P4122P4322), (P41212P43212)I422, I4122
Trigonal P3, (P31P32) R3, R32P312, (P3112P3212)P321, (P3121P3221)
Hexagonal P6, P63 ,(P61P65 ), (P62P64)P622, P6322, (P6122P6522), (P6222P6422)
Cubic P23, P213, I23, I213 P432, P4232, F432,F4132, I432, I4132
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Space group determination
Vario
us jo
bs a
re la
unch
ed o
n di
ffere
ntcl
uste
r nod
es in
var
ious
spa
ce g
roup
s
T=0
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Space group determination
T=0 + x
Vario
us jo
bs a
re la
unch
ed o
n di
ffere
ntcl
uste
r nod
es in
var
ious
spa
ce g
roup
s
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Space group determination
T=0 + x
Vario
us jo
bs a
re la
unch
ed o
n di
ffere
ntcl
uste
r nod
es in
var
ious
spa
ce g
roup
s
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Space group determination
T=0 + x
Vario
us jo
bs a
re la
unch
ed o
n di
ffere
ntcl
uste
r nod
es in
var
ious
spa
ce g
roup
s
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
How long Auto-Rickshaw takes to complete the job
Residues/ASU(Average)
Mol./ASU Beamline version(approx. time)
Advanced version(approx. time)
500 2 15 min 180 min
2.0 GHz , 64 CPU cluster
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Terminate data collection earlier
Collect more data than planned
Get direct answer to the questions:
Did my experiment work?
Do I get a structure out of it now?
Is Auto-Rickshaw printing out for me the Methods section?
Impact on Beamline Users
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Example-1:Fewer Data Sufficient• A protein 593 residues long (solvent content 78%)• Diffraction to a resolution of 2.55 Å• Nine Se heavy atoms• Space group P6422and
X-ray data data collection Auto-Rickshaw(CCP4+SHELXD+SHELXE+ARP/wARP)
• peak (360°) - finished and processed • inf (180°) - initiated and halted
Energy of incoming X-rays
X-ra
y an
omal
ous
sign
alInf
Peak
Rem
solved in 8 min (369 α-helical atom)130 residues traced at cycle 0 (30 min)561 residues traced after 2.5 hours
SAD
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Energy of incoming X-rays
X-ra
y an
omal
ous
sign
alInf
Peak
Rem
Example-2:More Data NeededSAD
• A protein 250 residues long (48% solvent content)• Diffraction to a resolution of 3 Å• One Pt heavy atom expectedbut…• SAD at peak - no solution• 2W-MAD at peak and inflection - no solution • 3W-MAD at peak, inflection and remote - no solution• SAD on 16-fold redundancy (peak+remote) YES !!!
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Combination of molecular replacement (MR)
and experimental (SAD)
phasing
-
Data reduction
Location of Heavy atoms
Heavy atom refinementand
Phase calculation
Solvent flattening&
NCS averaging
Model building
Model refinement
Expe
rimen
tal P
hasi
ng
Data reduction
Rotation function
Translation function
Rigid body refinement
Model building
Model refinement
MR
Pha
sing
• SAD• 2W-MAD• 3W-MAD• 4W-MAD• SIRAS• RIP
MR
Sear
ch M
odel
Hea
vy a
tom
der
ivat
ives
Steps in structure determinationAuto-Rickshaw
http://www.embl-hamburg.de/Auto-Rickshaw
A platform for automated crystal structuredetermination
MRSAD
Phased MR
MRRIP
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
The Search Model
Molecular replacement only works if you have a good search model.HOMOLOGY• The higher the sequence identity the easier MR.• There is a rough inverse correlation between the rms deviation of atomic
positions and the percentage sequence identity. • It is not possible to give an minimum value of homology at which MR will
work. Below 30% it is usually difficult but it can be challenging with a 100% identical molecule.
COMPLETENESS• The search model should represent a significant fraction of the unknown
structure. If it only represents 10-20% of the unknown structure MR will be difficult.
FLEXIBILITY• Some proteins (for instance antibodies) have domains that can adopt
different relative orientations with respect to each other. • MR will be difficult if the search model does not have the same relative
orientation of these domains
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
• AMORE (Navaza,1994)• MOLREP (Vagin & Teplyakov, 1997)• PHASER (Read., 2001)
• MRBUMP (Keegan & Winn, 2008)• BALBES (Long et al., 2008)
The MR programs and pipelines
The software pipeline makes several decisions concerning (i) truncation of the model in uncertain parts; (ii) the actual protocol for sequence alignment and homology modelling; and (iii) the choice of the MR software, the consensus approach is to derive a variety of models and try MR for all of them one by one
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Molecular Replacement
• In principle, accurate phases can be obtained in a matter of hours or even minutes.
• In practice, however, sometimes structure determination by MR is not straightforward.
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Why ?
• One major problem is that maps calculated from MR phases inevitably retain some memory of the search model (model bias)
• The more similar the phasing model to the unknown is, the less biased the calculated map will be.
σA -weighted
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Reduction of model bias
• Repeated cycling of real-space and reciprocal-space refinement • Density modification and NCS-averaging• Free atom modelling, refinement and model building as implemented in
ARP/wARP (Perrakis et al., 1999)• SHAKE and wARP (Reddy et al., 2003)• An ML-based reciprocal-space density-modification method (Prime &
Switch ) (Terwilliger et al., 2004)
• Omission of parts of the model (Bhat et al, 1988)• Simulated-annealing OMIT maps (Bruenger 1998)• Iterative-build OMIT maps (Terwilliger et al., 2008)
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
The sub-structure model
Experimental phasing only works if you have a substructure model.
• A complete and accurate substructure will result in better phase estimates than partial and/or inaccurate substructure.
• High quality data is needed to calculate structure-factor amplitudes of the anomalous scatterers and to locate the anomalous scatters.
• The resolution cutoff is an important parameter to obtain the best possible substructure from a given set of diffraction data.
• Radiation damage, low redundant and large substructure can be challenging to solve a substructure.
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
MRSAD
• An attractive feature of MRSAD strategy is that the anomalous scatterer substructure can be quickly determined from the MR solution.
• SAD phases are virtually independent of the MR phases, MRSAD provides an experimental method to overcome the model bias inherent to MR
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Basic difference in SAD phasing and MRSAD phasing
• Substructure determination
• Hand determination• Heavy atom refinement
and phase calculation• Density modification,
NCS-averaging and phase extension
• Model building
• MR + Anomalous difference Fourier
• No need of hand determination• Heavy atom refinement and
phase calculation• Density modification,
NCS-averaging and phase extension
• Model building
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Validation of MR
• Collection of Anomalous data Model refinement
Calculation of Anomalous difference Fourier or LLG map
MR
Density modification
0
0,5
1
1,5
2
4000 7200 10400 13600 16800 20000 23200
ENERGY
f“
1.54 Å 1.0 – 0.8 Å
1pjk
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Use of anomalous peaks in phase improvement
Phase combination of resultant phases, density
modification and/orNCS-averaging
Calculation of Anomalous difference Fourier or LLG map
Refinement of anomalous peaks (above 5σ) at max resolution
Dual fragment phasing usingthe partially refined/built model
and anomalous peaks
Model building
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Phase combination of resultant phases, density
modification and/orNCS-averaging
Model refinement
Calculation of Anomalous difference Fourier or LLG map
Refinement of anomalous peaks (above 5σ) at max resolution
Dual fragment phasing usingthe partially refined/built model
and anomalous peaks
Model building
MR
Density modification
Combination of MR and SAD phasing
Phase improvement of poor MR or SAD phaseand model completion
Validation ofMR solution
Poor MR PhasePoor SAD Phase
Iterative model completion and
refinement
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
αSAD αMOD
XP,SAD
αC,SAD
αCOMB
XH
∆FO
DM
EXIT
Panjikar, S., Parthasarathy, V., Lamzin, V. S., Weiss, M. S. & Tucker, P. A. . On thecombination of molecular replacement and single anomalous diffraction phasing forautomated structure determination Acta Cryst. D 2009 D65, 1089-1097
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
αSAD αMOD
XP,SAD
αC,SAD
αCOMB
XH
∆FO
∆FO
DM
EXIT
Panjikar, S., Parthasarathy, V., Lamzin, V. S., Weiss, M. S. & Tucker, P. A. . On thecombination of molecular replacement and single anomalous diffraction phasing forautomated structure determination Acta Cryst. D 2009 D65, 1089-1097
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
αSAD αMOD
XP,SAD
αC,SAD
αCOMB
XH
∆FO
αC,MR
XP,MR
FO
∆FO
DM
EXIT
MR + refine
Panjikar, S., Parthasarathy, V., Lamzin, V. S., Weiss, M. S. & Tucker, P. A. . On thecombination of molecular replacement and single anomalous diffraction phasing forautomated structure determination Acta Cryst. D 2009 D65, 1089-1097
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
MRSAD: combination of MR and experimental phasing
Phase combination of resultant phases and density
modification
Model refinement
Calculation of Anomalous difference Fourier map
Refinement of anomalous peaks (above 5σ)
at maximum resolution
Dual fragment phasing usingthe partial refined model and
anomalous peaks
Model building
MRSoftware pipeline for MRSAD
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
2G4L hydroxynitrile lyase 257 1 18 1.84 C2221 0.022 0.067 12 0.83 -----
2G4J Glucose isomerase 386 1 11 1.85 I222 0.036 0.141 13 0.86 ----- 1.80
2G4O LeuB 355 4 32 2.00 P212121 0.019 0.06 12 0.90 -----
2G4R MogA 156 3 03 1.92 P21 0.019 0.07 06 0.91 -----
2G4W Ribonuclease A (C2) 128 2 18 1.84 C2 0.022 0.063 06 0.97 ----- 1.00
2G4I Concanavalin A 237 1 07 2.40 I222 0.037 0.104 11 1.05 ----- 1.60
2G4K hARH3 256 1 19 1.82 C2221 0.019 0.069 12 1.06 ----- 2.20
2G51 Trypsin (p1) 240 1 10 1.84 P1 0.02 0.06 03 1.09 -----
2G4X Ribonuclease A (P3221) 128 1 18 1.95 P3221 0.019 0.053 18 1.10 ----- 1.40
2G4N α-lactalbumin 122 6 54 2.30 P21212 0.03 0.071 12 1.12 -----
2G4Y Thaumatin 207 1 16 1.98 P41212 0.013 0.064 25 1.12 0.682
2G4T PPE (Na) 240 1 13 1.84 P212121 0.015 0.043 12 1.15 0.549
2G4V Proteinase K 279 1 13 2.14 P43212 0.018 0.045 26 1.18 0.559
2G4S NBR1PB1 85 1 05 2.15 P6322 0.013 0.053 34 1.18 0.558
2G52 Trypsin (P21) 240 1 12 1.84 P21 0.033 0.076 06 1.22 ----- 1.80
2G4U PPE (Ca) 240 1 13 1.84 P212121 0.033 0.123 13 1.38 0.667
2G55 Trypsin (P3121) 240 1 10 1.82 P3121 0.019 0.086 10 1.46 0.708
2G4Z Thermolysin 316 1 06 1.98 P6122 0.018 0.067 38 1.64 0.655
2G4Q Lysozyme at pH 8.0 129 1 15 1.84 P43212 0.024 0.031 23 1.73 0.675
2G4P Lysozyme at pH 4.5 129 1 17 1.84 P43212 0.022 0.04 24 2.00 0.645
2G4M Insulin 51 1 06 1.80 I213 0.019 0.057 35 2.00 0.788
Mueller-D
ieckmann et al., 2007
Long wavelength (2.0 Å) dataset Longwavelength dataset with PDB code
Total residue per molecule = 110 - 755Number of mol. in asu = 1 - 4
Total number of Se in asu = 2 - 32Resolution =3.00 - 1.80 Å
Space group
Ranom=0.013 - 0.036Rmerge=0.031 - 0.141Multiplicity= 3 - 38
Ranom/Rpim= 0.83 - 2.00
Data quality
SAD
MR
SAD
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
2hh6 116 1 06 2.04 P6522 2o4t 0.44 1.04/072 0.056 0.10 7.5 1.43 N Y
1vky 563 1 06 3.00 I222 1yy3 0.49 1.84/247 0.086 0.126 3.5 1.10 Y N
2gi3 452 1 13 1.80 P3221 2g5i 0.50 1.34/403 0.058 0.112 6.3 1.19 Y N
2hxv 354 1 05 2.80 I222 2d5n 0.41 2.32/316 0.057 0.078 4.0 1.27 Y N
2o08 193 2 08 2.90 C2221 2ogi 0.42 1.30/180 0.058 0.07 3.7 1.36 Y Y
1vmf 407 1 05 1.90 P212121 2cu5 0.42 1.09/124 0.107 0.134 4.1 1.41 Y N
1zbt 320 1 08 2.40 P43212 2b3t 0.49 2.43/224 0.043 0.071 6.7 1.45 Y Y
1vmi 329 1 08 2.32 P6322 1xco 0.42 1.76/311 0.046 0.062 5.0 1.48 Y Y
2f4l 276 4 28 2.50 P212121 2ii1 0.36 1.03/274 0.075 0.084 3.8 1.49 Y N
2fvg 311 1 06 2.50 P212121 1vhe 0.37 1.70/227 0.073 0.081 3.8 1.51 Y Y
1vjo 755 1 08 2.00 P212121 2ch2 0.43 1.15/371 0.061 0.064 4.1 1.68 Y Y
1vjr 269 1 06 2.40 P41212 1zjj 0.40 1.55/240 0.076 0.072 4.6 2.00 Y Y
1vkn 338 4 32 2.45 P21 1xyg 0.46 1.18/316 0.069 0.065 6.1 2.40 Y Y
MRSAD examples on JCSG SeMet data
Search model with PDB codesequence identity= 35 - 51%rmsd after optimal overlap = 1.2 - 2.4 Å
JCSG dataset with PDB code Total residue per molecule = 110 - 755
Number of mol. in asu = 1 - 4Total number of Se in asu = 2 - 32
Resolution =3.00 - 1.80 ÅSpace group
Ranom=0.032 - 0.107Rmerge=0.064 - 0.134
Multiplicity=3.4 - 7.5Ranom/Rpim=0.67 - 2.40
MR
SAD
SAD
Dataset Search model quality Data quality
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Phase combination of resultant phases, density
modification and/orNCS-averaging
Model refinement
Calculation of Anomalous difference Fourier or LLG map
Refinement of anomalous peaks (above 5σ) at max resolution
Dual fragment phasing usingthe partially refined/built model
and anomalous peaks
Model building
MR
Density modification
Initial partial model built from SAD phasing protocol
Residue per monomer: 261Molecules in ASU: 4Max Resolution: 2.3 ÅSpace group: P21
Electron density after SAD phasing
Part of the complete structure, shown in experimental map
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Evolution of substructure, electron density and model
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Phase combination of resultant phases, density
modification and/orNCS-averaging
Model refinement
Calculation of Anomalous difference Fourier or LLG map
Refinement of anomalous peaks (above 5σ) at max resolution
Dual fragment phasing usingthe partially refined/built model
and anomalous peaks
Model building
MR
Density modification
Initial partial model built from SAD phasing protocol
Residue per monomer: 145Molecules in ASU: 4Max Resolution: 2.75 ÅSpace group: P43212
Initial partial model built from SAD phasing
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Evolution of the model at low resolution in MRSAD
51.9/55.5
47.6/57.6 47.6/55.8 46.5/53.9 43.4/53.6
32.7/41.5
29.6/34.6
http://cluster.embl-hamburg.de/lresult/LV1SvIl/LOG/viewlog/result.html�
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Model building module
RESOLVE
ARP/wARP
SHELXE
RESOLVE
BUCCANEER
SHELXE
Best phases after density modification
Poly-ala
Sequence docking andmodel expension
Iterative refinement, model building and sequence docking
SA
D fu
nctio
n
High resolutionbetter than 2.6 Å
Low resolution2.6 - 3.2 Å
Refinement using REFMAC5
Model Model
Refinement using REFMAC5
d50 = dmin[sc-1 - 1]1/3
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Parameters to gauge the map improvement and model completion
• Standard deviation of the local r.m.s. of the electron density map after density modification
• The Rfree value from the refinement of the model • Fraction of the model built• Fraction of sequence docked• Absolute peak heights in the anomalous difference Fourier
map
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Density modification
Phase combination of resultant phases, density
modification and/orNCS-averaging
Model refinement
Calculation of Anomalous/isomorphous
difference Fourier or LLG map
Refinement of anomalous/isomorphous peaks (above 5σ) at max resolution
Dual fragment phasing usingthe partialy refined/built model
and anomalous/isomorhous peaks
Model building
MR
SAD
SIRAS
2W-MAD
3W-MAD
4W-MAD
Hea
vy a
tom
refin
emen
t
Substructure determinationValidation of MR
Ligand evaluation
Phase improvementand modelcompletion
RIP/SIR
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
• MRSAD: (1) A means of completing your model(2) Overcome model bias from MR solution(3) Use long wavelength X-ray data for MR
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Native data collection
0
0,5
1
1,5
2
4000 7200 10400 13600 16800 20000 23200
ENERGY
f“
1.54 Å 1.0 – 0.8 Å
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Phasing at short wavelength
Heavy atoms/ASU
2 Zn + 12S
Wavelength (Å) 0.8
f” 1.76 (Zn) 0.15(S)
Bijvoet ratio 1.14
Space group P21Redundancy 3.7
Resolution (Å) 20-1.66
Completeness (%) 97.4 (94.3)
I/σ(I) 14.2(2.8)
Rmerge (%)b 3.7 (37.0)
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Phasing at short wavelength
Heavy atoms/ASU 6S
Wavelength (Å) 1.00
f” 0.24(S)
Bijvoet ratio 0.66
Space group I213
Redundancy 22.6
Resolution (Å) 20-1.6
Completeness (%) 98.2(100.0)
I/σ(I) 42.0 (13.0)
Rmerge (%)b 3.2 (12.2)
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Phasing at short wavelength
Heavy atoms/ASU 17S
Wavelength (Å) 0.9
f” 0.20(S)
Bijvoet ratio (%) 0.45
Space group P41212
Redundancy 7.6
Resolution (Å) 20-1.4
Completeness (%) 94 (68)
I/σ(I) 33.3(9.8)
Rmerge (%)b 3.9(13.7)
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Exploiting Phase Information
We may also find ourselves in a situation where we have experimental phases (from a heavy atom derivative) as well as a model.
Phase information can, in fact, be exploited to help to solve the MR problem.
This is particularly useful when we have a molecular replacement model that is too poor to find a solution and phases that are too weak to provide an interpretable map.
By combining molecular replacement with experimental phases, the two together may be enough to succeed.
This technique of combining experimental phases with molecularreplacement methods is called phased molecular replacement.
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Phased Molecular Replacement
Domain rotationfunction
FTFA
Orientation (αA, βA, γA)
FT
F(αA, βA, γA)Phased translationfunction
(TxA,TyA, TzA)
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
An Example
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Another example
Phased MR model is useful for interpretability of the poor experimental mapand this can be also useful in phase enhancement
90°
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
New emerging phasing method: RIP X-ray Vs. UV
X-ray UV
Nano & Ravelli (2006)
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Data collection for UV-RIP phasing
X-ray data collection before UV
UV exposure (10 – 30 min)
X-ray data collection after UV
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
CYS-158: 27.6 CYS-174: 14.9
CYS-194: 24.7CYS-127: 21.0
CYS-46:20.1CYS-30:23.8
CYS-184:14.4CYS-214: 8.8
TRP-164
TRP-12
TRP-132
TRP-39
TRP-26TRP-83
TRP-232
Elastase5 min series(2)
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
TRP-CYS distance
CYS-158: 27.6 CYS-174: 14.9
CYS-194: 24.7CYS-127: 21.0
CYS:46:20.1CYS:30:23.8
CYS-184:14.4CYS-214: 8.8
12-TRP 20.0 09.2 19.3 19.6
26-TRP 33.8 31.0 14.9 24.8
39-TRP 28.6 25.1 17.0 29.4
83-TRP 15.9 20.9 09.8 16.9
132-TRP 24.6 18.5 12.2 16.6
164-TRP 08.0 17.5 16.1 11.0
232-TRP 21.3 20.4 13.6 24.7
Elastase5 min series
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
UV radiation does not only break disulphide bridges but also damage seleno-methionine and covalently bound heavy atoms.
UV radiation induced phasing with or without
S/MAD, S/MIR(A)S and MR can provide additional or
alternative phasing information
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
UV-RIP phasing from Seleno-Met protein crystalFAE
H32
ChSynt
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Auto-Rickshaw server
Commandline
Web-browser
BeamlineAdvanced
Version
Necessary Input required
Overview of Auto-Rickshaw
AR server is available to world wide crystallographic community
QuickLongerSAD
MR
MRSIRAS
SIRASMRMAD
MAD
MRRIP
RIP
S-SADMRSAD
No. of residues per monomer
No. of expected heavy atom sites per monomer
No. of monomers per a.u.
Space group
X-ray data
Sequence file (optional)
Model file PDB-format (optional) Email address
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Input Required
integrated and scaled X-ray data files
formats Extension SCALEPACK : H K L I+ SIGI+ I- SIGI- ( .sca )
MTZ : H K L I+ SIGI+ I- SIGI- ( .mtz )
XDS : XDS-ASCII.HKL ( .HKL ) or ( .hkl )
XSCALE : ( .HKL ) or ( .hkl )
D*TREK (.REF) or (.ref)
sequence file (if available)
free format ( .pir ) or ( .seq) or ( .txt)
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Beamline version(Phasing, density modification and helix/β-strand recognition)
Advanced version(Phasing, density modification and modelbuilding)
SAD, 2W-MAD, 3W-MAD, 4W-MAD, SIRAS
SAD, S-SAD,
2W-MAD, 3W-MAD, 4W-MAD,
SIRAS,
RIP
MR, MRSAD, MRSIRAS,MR-2W-MAD, MR-3W-MAD,MR-4W-MAD, MR-RIP
Web-browser Interface for job submission
Quick
longer
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Strategy for data evaluation at third generation source
Data collection
Data Processing
Data evaluation
Group-1
Group-2Group-3
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Text for materials and methods with acknowledgement
An automatic output for the materials and methods
It includes also the references for all the programs whichare used for the structure solution.
It guides the user how to acknowledge the pipeline along with the software used in the pipeline.
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
• More than 1500 novel structures solved
• Largest structure solved : asymmetric unit contains 7,500 residues
• Largest substructure solved: 140 heavy atoms sites in asymmetric unit
Status and Highlights
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Auto-Rickshaw
Reduction of the number of input parameters Addition of other phasing protocols (i.e. MIR) Parallel computing at various steps of structure determination Scanning and improvement of decision making parameters• Remote connection with data collection/processing software ( i.e. XIA2 and DNA)
• Auto-Rickshaw provides results in almost real time for data collection at home sources or second generation synchrotrons. Accelerating it further for third generation sources is a great challenge.
The Future
-
Total registered institutes : 283 Total users from registered institutes : 754
57
7
15
45
18
39
241
24
47
13
15
38
20
1
8
5
26
7
1
3
Geographic distribution of AR users
1
3
19
7
3
1
3
310
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
More infos:
The Auto-Rickshaw Web Site:
http://www.embl-hamburg.de/Auto-Rickshaw
Auto-Rickshaw
Auto-Rickshaw and EMBL
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Acknowledgements
Santosh PanjikarVenkataraman Parthasarathy
Manfred S. WeissVictor S. LamzinPaul A. Tucker
All the developers of the various computer programs for their kind permissionto use their software in this pipeline (Martyn Winn, Kevin Cowtan, Qang Hao,George Sheldrick, Gerard Bricogne, Clemens Vonrhein, Navraj Pannu, TomTerwilliger, Axel Brunger, Gerard Kleywegt, Alwyn Jones, G. D. Smith, VictorLamzin, Anastassis Perrakis, Randy Read, Garib Murshudov, Alexi Vagin, Hi-Fu Fan, Thomas Schneider).
Hamburg beamline/Auto-Rickshaw users and JCSG for the provision of data
Thank you for your attention
Kay Diederichs for testing DPS2AR module
Daniele De Sanctis (UV-RIP )�
-
Macromolecular Crystallography School 2010, Madrid SP :Auto-Rickshaw
Tutorial
http://www.embl-hamburg.de/~panjikar/MAD/
http://www.embl-hamburg.de/~panjikar/MAD/�http://www.embl-hamburg.de/~panjikar/MAD/�http://www.embl-hamburg.de/~panjikar/MAD/�
Fundamental of�Automated Structure Determination From Crystal to StructureFrom Crystal to StructureFrom Crystal to StructureFrom Crystal to StructureFrom Crystal to StructureFrom Crystal to StructurePhasing in Macromolecular CrystallographyPhase Determination Methods��Steps in structure determination (Experimental phasing)�Número de diapositiva 12Steps in structure determination (Experimental phasing)�Primary aimUltimate aimThe Auto-Rickshaw PhilosophyThe Auto-Rickshaw ApproachThe Auto-Rickshaw ApproachThe Auto-Rickshaw ApproachNúmero de diapositiva 20Decision making processNúmero de diapositiva 22Número de diapositiva 23The Auto-Rickshaw System for experimental phasingNúmero de diapositiva 25Número de diapositiva 26Decision makingResolution cut offResolution cut offResolution cut offDecision makingWhen to terminate the SHELXD trialsSubstructure determination using SHELXD in Auto-RickshawSubstructure determination using SHELXD in AutoRickshawDecision makingChanging hand of the substructureChanging hand of the substructureChanging hand of the substructureEnantiomorph determinationDecision makingExample: which software to choose for Phase Calculation ?Example: which software to choose for Phase Calculation ?Determination of correct space groupSpace group determinationSpace group determinationSpace group determinationSpace group determinationHow long Auto-Rickshaw takes to complete the jobNúmero de diapositiva 49Example-1:Fewer Data SufficientNúmero de diapositiva 51Combination of �molecular replacement (MR) �and �experimental (SAD)�phasing���The Search ModelThe MR programs and pipelinesMolecular ReplacementWhy ? Reduction of model biasThe sub-structure modelMRSADBasic difference in SAD phasing and MRSAD phasingValidation of MRUse of anomalous peaks in phase improvementNúmero de diapositiva 64Número de diapositiva 65Número de diapositiva 66Número de diapositiva 67MRSAD: combination of MR and experimental phasingNúmero de diapositiva 69Número de diapositiva 70Número de diapositiva 71Evolution of substructure, electron density and modelNúmero de diapositiva 73Número de diapositiva 74Model building moduleParameters to gauge the map improvement and �model completion Número de diapositiva 77Número de diapositiva 78Native data collectionPhasing at short wavelengthPhasing at short wavelengthPhasing at short wavelengthNúmero de diapositiva 83Phased Molecular ReplacementNúmero de diapositiva 85Número de diapositiva 86New emerging phasing method: RIP �X-ray Vs. UV Data collection for UV-RIP phasingNúmero de diapositiva 89TRP-CYS distanceNúmero de diapositiva 91UV-RIP phasing from Seleno-Met protein crystalNúmero de diapositiva 93Número de diapositiva 94Número de diapositiva 95Input RequiredNúmero de diapositiva 97Strategy for data evaluation at third generation sourceText for materials and methods with acknowledgementNúmero de diapositiva 100Auto-RickshawNúmero de diapositiva 102Auto-RickshawAcknowledgementsTutorial