NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold...
Transcript of NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold...
![Page 1: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/1.jpg)
NYSBC-NCCAT 2020Single-particle cryo-EM Course
ValidationMethods
A blast from the past …
![Page 2: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/2.jpg)
The dark side of single-particle EM
The great thing about single-particle EM:Every data set and processing approach yields a 3D map !
The bad thing about single-particle EM:Every data set and processing approach yields a 3D map !
But is it correct ???
Particularly problematicfor low-resolution maps
![Page 3: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/3.jpg)
The issue: Structures of the IP3 receptoras determined by single-particle EM
Jiang et al.,2002
Serysheva et al.,2003
Jiang et al.,2003
Sato et al.,2004
![Page 4: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/4.jpg)
Grid
Specimenpreparation
2D images
Imaging
2D averages
Particle pickingAlignment2D classification
Initial 3D map
3D reconstruction
Final 3D map(s)
3D classificationRefinement
Cells
Protein
ExpressionPurification
Structure determination by single-particle EM
3D model(s)
Validation
OptimizationValidation
Image processing
Molecular biologyBiochemistry
Electron microscopy
Interpretation
ValidationDockingModel building
![Page 5: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/5.jpg)
Structure determination by single-particle EMCells
Protein
ExpressionPurification
Potential issues:
Heterogeneity– Compositional– Conformational
– Discrete states– Continuous movement
Effect of cross-linking
![Page 6: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/6.jpg)
Structure determination by single-particle EM
Potential issues with samples
If chemical fixation was used:Look at unfixed sample to assess effect of cross-linking
àAssess whether structure of cross-linked sample is meaningful
Before attempting structure determination –Understand and optimize your sample !
Prepare negatively stained specimens:Good contrast and preferred orientations
à Easy to assess heterogeneity
If particles look heterogeneous:Calculate class averages
à Assess type and degree of heterogeneityà Minimize heterogeneity by any means possible
![Page 7: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/7.jpg)
Effect of cross-linking:The HOPS tethering complex
Cross-linked
Bröcker et al. (2012)PNAS 109: 1991-1996
Native
Chou et al. (2016)NSMB 23: 761-763
C B
AH
![Page 8: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/8.jpg)
Structure determination by single-particle EM
Grid
Specimenpreparation
Cells
Protein
ExpressionPurification
Potential issues:
– No particles
– Preferred orientations
![Page 9: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/9.jpg)
Structure determination by single-particle EM
Potential issues with grids
No particles (particles bind to carbon and avoid holes)– Increase protein concentration– Double blotting– Use thin support film (carbon or graphene oxide)– Use different grids, e.g., PEG-treated or gold grids
Preferred orientation (particles align at air/water interface)
– Use low concentration of detergent (changes surface tension)– Use thin carbon film (commonly used for ribosome samples)– Use gold grids
Lack of views will result in:– non-isotropic resolution of the density map– can potentially lead to an incorrect density map
Different sample preparation approach (e.g., Spotiton)Collect images from tilted specimens
![Page 10: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/10.jpg)
Preferred orientations: Pex1/6 complexWithout detergent
![Page 11: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/11.jpg)
Preferred orientations: Pex1/6 complexWith detergent
![Page 12: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/12.jpg)
Structure determination by single-particle EM
Grid
Specimenpreparation
2D images
Imaging
Cells
Protein
ExpressionPurification
Potential issues:
– Low contrast
– Beam damage
![Page 13: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/13.jpg)
Structure determination by single-particle EM
Potential issues with images
Poor electron scatteringà high electron dose
Beam sensitivityà low electron dose
àPoor SNR can be fixedby averaging
àLoss of informationcannot be fixed
àElectron micrographs recorded with low electron dosesàParticles hard too see, especially small ones
Problem fixed by DDD cameras
àCollect long moviesàAdd frames with resolution filter
(removes damaged high-resolution informationretains low-resolution information for good SNR)
![Page 14: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/14.jpg)
Structure determination by single-particle EM
Grid
Specimenpreparation
2D images
Imaging
2D averages
Particle pickingAlignment2D classification
Cells
Protein
ExpressionPurification
Particle picking:– Model/reference bias
2D classification:– Model/reference bias– Number of classes– Heterogeneous classes– Disappearing classes
Potential issues:
![Page 15: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/15.jpg)
Structure determination by single-particle EM
Potential issues with particle picking
Shatsky et al. (2009) J. Struct. Biol. 166: 67-78Henderson (2013) Proc. Natl. Acad. Sci. USA 110: 18037-18041
1,000 images containingpure white noise
Reference:Albert Einstein
![Page 16: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/16.jpg)
Structure determination by single-particle EM
Potential issues with particle picking
Average of 1,000 images containingpure white noise after alignment to an image of Albert Einstein
à Einstein from noise
Model/reference bias
Shatsky et al. (2009) J. Struct. Biol. 166: 67-78Henderson (2013) Proc. Natl. Acad. Sci. USA 110: 18037-18041
![Page 17: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/17.jpg)
Structure determination by single-particle EM
Potential issues with particle picking
Mao et al. (2013)PNAS 110: 12438-12443
Henderson (2013)PNAS 110: 18037-18041
HIV env trimer
b-galactosidase
![Page 18: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/18.jpg)
Structure determination by single-particle EM
Potential issues with particle picking
Henderson (2013)PNAS 110: 18037-18041
Mao et al. (2013)PNAS 110: 12438-12443
HIV env trimer
Using template matchingto pick particles from verynoisy images is dangerous
à Averages will end uplooking like templates usedfor particle picking
à Better to first pick imageswithout templates and useresulting averages as templates for re-picking
![Page 19: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/19.jpg)
Structure determination by single-particle EM
Grid
Specimenpreparation
2D images
Imaging
2D averages
Initial 3D map
3D reconstruction
Cells
Protein
ExpressionPurification
Particle pickingAlignment2D classification
Incorrect map
Because of:– Heterogeneous sample– Missing views– Incorrect solution
Potential issues:
![Page 20: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/20.jpg)
IMAGING
3D reconstructionof specimen
BACKPROJECTION
specimenat differenttilt angles
Different projection views
ASSIGN ORIENTATIONPARAMETERS x, y and F
Random conical tilt reconstruction
![Page 21: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/21.jpg)
x
y
g
a
b5 parametersto determine
Single particles in ice
![Page 22: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/22.jpg)
Angular reconstitution
Serysheva et al., 1995
2. add in further projections and keeprefining
1. choose 3 projection images that areperpendicular views of the particle(anchor set)
van Heel, 1987
![Page 23: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/23.jpg)
Chicken Slo2.2 in the absence of Na+
Class averages Initial model (obtained with VIPER)
VIPER
StochasticHill Climbing
(initially introducedin program SIMPLE)
![Page 24: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/24.jpg)
Systemetically generatedprojections from existing
3D reconstruction
Stack ofreference
projections
Experimentalprojection
max. CCFcoefficient
Refinedorientationparameters
Projectionmatching
Back-projection generatesnew 3D reconstruction
Angular refinement
Stack ofCCFs
SteepestDecent
![Page 25: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/25.jpg)
Systemetically generatedprojections from existing
3D reconstruction
Stack ofreference
projections
Experimentalprojection
FIRSTmax. CCFcoefficient
Refinedorientationparameters
Projectionmatching
Back-projection generatesnew 3D reconstruction
Angular refinement
Stack ofCCFs
Stochastic HillClimbing
![Page 26: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/26.jpg)
Structure determination by single-particle EM
Grid
Specimenpreparation
2D images
Imaging
2D averages
Initial 3D map
3D reconstruction
Final 3D map(s)
3D classificationRefinement
Cells
Protein
ExpressionPurification
Particle pickingAlignment2D classification
Potential issues:
Reference bias
Overfitting
Resolution assessment
![Page 27: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/27.jpg)
Structure determination by single-particle EM
Potential issues with density map
Shatsky et al. (2009) J. Struct. Biol. 166: 67-78Henderson (2013) Proc. Natl. Acad. Sci. USA 110: 18037-18041
Average of 1,000 images containingpure white noise after alignment to an image of Albert Einstein
à Einstein from noise
Model/reference bias
![Page 28: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/28.jpg)
Systemetically generatedprojections from existing
3D reconstruction
Stack ofreference
projections
Experimentalprojection
max. CCFcoefficient
Refinedorientationalparameters
Stack ofCCFs
Projectionmatching
Back-projection generatesnew 3D reconstruction
Angular refinement
SteepestDecent
![Page 29: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/29.jpg)
Structure determination by single-particle EM
Potential issues with density map
Shatsky et al. (2009) J. Struct. Biol. 166: 67-78Henderson (2013) Proc. Natl. Acad. Sci. USA 110: 18037-18041
Average of 1,000 images containingpure white noise after alignment to an image of Albert Einstein
à Einstein from noise
Model/reference bias
Over-fitting results in spurious high-resolution features due to alignmentof noise
![Page 30: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/30.jpg)
Resolution [Å]
Fou
rier
shel
l cor
rela
tion 1
0
Structure determination by single-particle EM
Resolution assessment
0.5
0.143
11 8.7
FSC = 0.143 Phase error = 60ºRosenthal & Henderson (2003) J. Mol. Biol. 333: 721-745
FSC = 0.5 Signal = NoiseBöttcher et al. (1997) Nature 386: 88-91
Maps have to beindependent !
![Page 31: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/31.jpg)
Structure determination by single-particle EM
Resolution assessment
Stack of particles(original orientation
parameters)
Stack of particles(refined orientation
parameters)
Referencemap
Refinedmap
Fourier shellcorrelation“Half maps”
Stack of particles(original orientation
parameters)
Refinedhalf maps
“Gold standard”Fourier shellcorrelation
Data set is split at the startà Truly independent half maps
Half maps not independent !
Referencemaps
Half stacks(refined orientation
parameters)Half stacks
![Page 32: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/32.jpg)
Structure determination by single-particle EM
Resolution assessment
Rosenthal & Rubinstein (2015) Curr. Opin. Struct. Biol. 34: 135-144
Rotavirus double-layered particle 20 Å
7 Å
3.8 Å
2.6 Å
> 20 Åprotein envelope
~ 9-10 Åa-helices
< 4.8 Åb-sheets
~ 4 Åbulky side chains
What should beresolved ?
![Page 33: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/33.jpg)
Structure determination by single-particle EM
Grid
Specimenpreparation
2D images
Imaging
2D averages
Initial 3D map
3D reconstruction
Final 3D map(s)
Cells
Protein
ExpressionPurification
Validation
Particle pickingAlignment2D classification
3D classificationRefinement
![Page 34: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/34.jpg)
The issue: Structures of the IP3 receptoras determined by single-particle EM
Jiang et al.,2002
Serysheva et al.,2003
Jiang et al.,2003
Sato et al.,2004
![Page 35: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/35.jpg)
Henderson et al. (2012) Structure 20: 205-214
Map validationMeeting of experts in 2010 to come up with standards for map validation
Outcome summarized in 2012:
![Page 36: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/36.jpg)
Henderson et al. (2012) Structure 20: 205-214
Map validation
– Compare reference-free averages with projections
![Page 37: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/37.jpg)
Map validationRe-projections and angular distribution
Anaphasepromotingcomplex
![Page 38: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/38.jpg)
Map validation
– Compare reference-free averages with projections
– Tilt-pair analysis
– only checks consistency of 3D map with 2D data– also check angle distribution
Henderson et al. (2012) Structure 20: 205-214
![Page 39: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/39.jpg)
Map validationTilt-pair analysis
–a
+a
Matching projections
Particle stack (–a)
Particle stack (+a)
Rosenthal & Rubinstein (2015) Curr. Opin. Struct. Biol. 34: 135-144
Rosenthal & Henderson (2003) J. Mol. Biol. 333: 721-745Henderson et al. (2011) J. Mol. Biol. 413: 1028-1046
à D angles
![Page 40: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/40.jpg)
Map validationTilt-pair analysis
Tilt-pair parameter plot
Rosenthal & Rubinstein (2015) Curr. Opin. Struct. Biol. 34: 135-144
Tilt-pair phase residual plot
![Page 41: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/41.jpg)
Map validationTilt-pair analysis
Henderson et al. (2011) J. Mol. Biol. 413: 1028-1046
– can be used to refine parameters used for orientation determinationà can thus be used to improve the map
– allows determination of handedness
– determines whether overall 3D map is correct at 15-20 Å resolution(but not high-resolution features)
– validates orientation parameters(but not microscope parameters, i.e., defocus, magnification)
“If less than 60% of particles show a single cluster, the basisfor poor orientation parameters should be investigated”
![Page 42: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/42.jpg)
Map validationTilt-pair analysis
Tilt-pair alignment test
angular errors for determination of the tilt transformation of each particle pair expected for random orientations
Rosenthal & Rubinstein (2015) Curr. Opin. Struct. Biol. 34: 135-144Baker et al. (2012) Proc. Natl Acad. Sci. USA 109: 11675-11680
Russo & Passmore (2014) J. Struct. Biol. 187: 112-118
![Page 43: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/43.jpg)
Map validationTilt-pair web server
Wasilewski & Rosenthal (2014) J. Struct. Biol. 186: 122-131
Input Output
![Page 44: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/44.jpg)
Map validationhttp://www.ebi.ac.uk/pdbe/emdb/validation/tiltpair/
![Page 45: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/45.jpg)
Henderson et al. (2012) Structure 20: 205-214
Map validation
– Compare reference-free averages with projections
– Tilt-pair analysis
– “Gold standard” FSC
– Randomize phases
– only checks consistency of 3D map with 2D data
– excellent, also establishes handedness
– not necessarily needed (but certainly not bad)
– also check angle distribution
![Page 46: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/46.jpg)
Map validationRandomize phases
Chen et al. (2013) Ultramicroscopy 135: 24-35
l Do single-particle reconstruction / refinement
l Determine resolution (FSC)
l Take raw data, randomize phases beyond which FSCT falls below a threshold (75 or 80%)
l Redo the same analysis and recalculate FSC curve
l Any signal in region of randomized phases indicates issues with noise alignment in that region
l Can be implemented in any package
Rosenthal & Rubinstein (2015) Curr. Opin. Struct. Biol. 34: 135-144
![Page 47: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/47.jpg)
Map validationRandomize phases
Rosenthal & Rubinstein (2015) Curr. Opin. Struct. Biol. 34: 135-144Chen et al. (2013) Ultramicroscopy 135: 24-35
FSC signal due to over-fitting (noise)FSC signal due to true structural information
![Page 48: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/48.jpg)
Henderson et al. (2012) Structure 20: 205-214
Map validation
– Compare reference-free averages with projections
– Tilt-pair analysis
– “Gold standard” FSC
– Randomize phases
– only checks consistency of 3D map with 2D data
– excellent, also establishes handedness
– not necessarily (but certainly not bad)
– excellent, but not commonly used
– also check angle distribution
– Appearance of expected secondary structure elements
![Page 49: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/49.jpg)
Samso et al. (2009) PLoS Biol. 7: e1000085
Map validationExpected secondary structure
Ryanodine receptor 1at 10.2 Å resolution
![Page 50: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/50.jpg)
Henderson et al. (2012) Structure 20: 205-214
Map validation
– Compare reference-free averages with projections
– Tilt-pair analysis
– “Gold standard” FSC
– Randomize phases
– Appearance of expected secondary structure elements
– only checks consistency of 3D map with 2D data
– excellent, also establishes handedness
– not necessarily (but certainly not bad)
– excellent, but not commonly used
– Evaluate with published information
– also check angle distribution
![Page 51: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/51.jpg)
Map validationEvaluation with published information
Anaphase promoting complex
![Page 52: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/52.jpg)
Henderson et al. (2012) Structure 20: 205-214
Map validation
– Compare reference-free averages with projections
– Tilt-pair analysis
– “Gold standard” FSC
– Randomize phases
– Appearance of expected secondary structure elements
– only checks consistency of 3D map with 2D data
– excellent, also establishes handedness
– not necessarily (but certainly not bad)
– excellent, but not commonly used
– Evaluate with published information
– also check angle distribution
– Dock known atomic structures into map
– yeast two-hybrid analysis– pull-down experiments– cross-link mass spectrometry
![Page 53: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/53.jpg)
mGluR1Kunishima et al. 2000
(K. Morikawa)
KcsADoyle et al. 1998(R. MacKinnon)
GluR2Armstrong et al. 2000
(E. Gouaux)
Map validationDocking of atomic models
![Page 54: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/54.jpg)
Map validationDocking of atomic models
![Page 55: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/55.jpg)
Map validationDocking of atomic models
![Page 56: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/56.jpg)
Native AMPA-R
Nakagawa et al. (2006)Biol. Chem. 387: 179-187
Sobolevsky et al. (2009)Nature 462: 745-758
Engineered GluA2
Tichelaar et al. (2004)JMB 344: 435-442
Map validationDocking of atomic models
![Page 57: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/57.jpg)
Jiang et al.,2002
Serysheva et al.,2003
Jiang et al.,2003
Sato et al.,2004
Map validation – IP3 receptorDifferent maps of the IP3 receptor
![Page 58: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/58.jpg)
Ludtke et al. (2011) Structure 19: 1192-1199
Map validation – IP3 receptorNew density map in 2011 at 11 Å resolution
![Page 59: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/59.jpg)
Kir2.2
Map validation – IP3 receptorExpected secondary structure elements
Ludtke et al. (2011) Structure 19: 1192-1199
![Page 60: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/60.jpg)
Murray et al. (2013) Structure 21: 900-909
Map validation – IP3 receptorComparison of reference-free averages with projections
A: Map projectionB: Reference-based class averageC: Reference-free class averageD: Selected particles
![Page 61: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/61.jpg)
Murray et al. (2013) Structure 21: 900-909
Map validation – IP3 receptorTilt pair test
Ludtke et al.2011
Serysheva et al.2003
Sato et al.2004
Jiang et al.2002
![Page 62: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/62.jpg)
Murray et al. (2013)Structure 21: 900-909
Map validation – IP3 receptorComparison of maps from different programs
![Page 63: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/63.jpg)
Fan et al. (2015) Nature 527: 336-341
Map validation – IP3 receptor4.7 Å resolution structure (2015)
![Page 64: NYSBC Course 2020 new · l Take raw data, randomize phases beyond which FSCTfalls below a threshold (75 or 80%) l Redo the same analysis and recalculate FSC curve l Any signal in](https://reader034.fdocuments.in/reader034/viewer/2022042219/5ec4e785d39111156f60e4b9/html5/thumbnails/64.jpg)
Map validation – IP3 receptor3.5 Å resolution structure (2018)
IP3R – no ligands
Paknejad & Hite (2018) Nat. Struct. Mol. Biol. 25: 660-668
Two IP3-bound conformations
IP3 class 1
IP3 class 2
Ensemble of IP3-bound conformations
IP3 class 1 IP3 class 2
* * *
*
**
*
**
*
Ca2+-bound conformation
No ligands Ca2+-bound