3D Reconstruction of Icosahedral Particles. NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

Transcript

1 3D Reconstruction of Icosahedral Particles NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

2 3D Reconstruction of Icosahedral Particles Outline - Background - References; examples; etc. - Symmetry - Icosahedral (532) point group symmetry - Triangulation symmetry ΞΨΖ ϑουρναλ ϖιρυσ χαυσεσοφ µυχη Digitize Micrograph Graphics Box G y x Determine Ιντερπρετ Ρεφινε Ιντερπαρτιχλε ετεχτ Νορµαλιζε θ Ξ Ψ Ζ φ ω Β ξψ Παρτιχλε Χοµπυτε αχκ Πυβλιση et ισπλαψ/ινσπεχτ/αναλψζε Ρανδ εφινε Ρεγυλαρ initial µορε out προϕεχτ παρτιχλε ιµαγε θ,φ and παρτιχλ θ,φ,ω,ξ,ψ στρυχτυρε 3 estimate ορ ιµαγε Ρεφινε ϖαλυεσ αρτιφαχτσ γριδ ρεχονστρυχτιον float περιση 3 οριεντατιον εοριγινσ µεανσ/ϖαριανχεσ στρυχτυρε οφ ιµαγεσ οριγινσ παραµετερσ θ,φ of (αστιγµατισµ, ρεφινεµεντ Ιχοσαηεδραλ ασψµµετριχ υνιτ: παιν?αδδ συφφερινγ ιν µεν, ςιρυσ ωοµεν, ανδστρυχτυρε χηιλδρεν... Ινχρεασε micrograph display individual particle Χορρελατε δριφτ, Ρεµοϖε Ιµποσε (Χροσσ χορρελατε ετχ.) βαχκγρουνδ 3 centers ρεσολυτιον particle ανδ ωιτη στρυχτυρε 532 εστιµατε σψµµετρψ οτηερ (x,y γραδιεντσ (ε.γ. ιµαγεσ ρεσολυτιον βιοχηεµιχαλ, ωιτη ιν ετχ.) δατα οριεντατιον (Χοµπαρε ϖαλυεσ ιν ιµαγεσ ασψµµετριχ ωιτη προϕεχτεδ υνιτ Σηαδεδ τριανγλε βουνδεδ βψ 2 φολδ, images origins) Φουριερ προϕεχτεδ τρανσφορµσ ϖιεωσ οφ 3 ρεχονστρυχτιον) ανγλεσ ϖιεωσ (θ,φ,ω) οφ 3 ρεχονστρυχτιον) 3 φολδ, ανδ τωο αδϕαχεντ 5 φολδ αξεσ - Typical procedure (flow chart) - Digitization and boxing - Image preprocessing / CTF estimation - Initial particle orientation/origin search - Orientation/origin refinement - 3D reconstruction with CTF corrections - Validation (resolution assessment) - Current and future strategies NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

3 3D Reconstruction of Icosahedral Particles REFERENCES Crowther, R. A., Amos, L. A., Finch, J. T., DeRosier, D. J. and Klug, A. (1970) Three dimensional reconstructions of spherical viruses by Fourier synthesis from electron micrographs. Nature 226: First 3D reconstructions of negatively-stained, spherical viruses: Human wart virus Tomato bushy stunt 500 Å NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

4 3D Reconstruction of Icosahedral Particles REFERENCES Crowther, R. A., DeRosier, D. J. and Klug, A. (1970) The reconstruction of a three-dimensional structure from projections and its application to electron microscopy. Proc. Roy. Soc. Lond. A 317: Crowther, R. A. (1971) Procedures for three-dimensional reconstruction of spherical viruses by Fourier synthesis from electron micrographs. Phil. Trans. R. Soc. Lond. B. 261: General principles of 3DR method - Fourier-Bessel mathematics - Common lines NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

5 3D Reconstruction of Icosahedral Particles REFERENCES - Reference list available as handout - For die-hards: Baker, T. S., N. H. Olson, and S. D. Fuller (1999) Adding the third dimension to virus life cycles: Three-Dimensional reconstruction of icosahedral viruses from cryo-electron micrographs. Microbiol. Molec. Biol. Reviews 63: NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

6 Reovirus Rotavirus HSV NωV LA-1 NβV SpV-4 Polyoma φx174 CaMV FHV HRV Å HPV Polio Ross River CPMV CCMV B19

7 HSV NωV LA-1 NβV SpV-4 Polyoma φx174 PBCV-1 FHV HRV Å Polio CPMV CCMV B19

8 HSV NωV LA-1 NβV SpV-4 Polyoma φx174 PBCV-1 FHV HRV Å Polio CPMV CCMV B19

9 3D Reconstruction of Icosahedral Particles Outline - Background - References; examples; etc. - Symmetry - Icosahedral (532) point group symmetry - Triangulation symmetry - Typical procedure (flow chart) - Digitization and boxing - Image preprocessing / CTF estimation - Initial particle orientation/origin search - Orientation/origin refinement - 3D reconstruction with CTF corrections - Validation (resolution assessment) - Current and future strategies NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

10 3D Reconstruction of Icosahedral Particles Symmetry 1. Icosahedral (532) point group symmetry 2. Triangulation symmetry NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

11 Regular Polyhedra (Platonic Solids) There are just five platonic solids: From equilateral triangles you can make: with 3 faces at each vertex, a tetrahedron with 4 faces at each vertex, an octahedron with 5 faces at each vertex, an icosahedron From squares you can make: with 3 faces at each vertex, a cube From pentagons you can make: with 3 faces at each vertex, a dodecahedron

12 Icosahedral (532) Point Group Symmetry 12 vertices (5fold) NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

13 Icosahedral (532) Point Group Symmetry 12 vertices (5fold) 20 faces (3fold) NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

14 Icosahedral (532) Point Group Symmetry 12 vertices (5fold) 20 faces (330fold) edges (2fold) NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

15 Icosahedron Dodecahedron Different shapes, but both have 532 symmetry 12 vertices, 20 faces, 30 edges (6 5-folds, 10 3-folds, 15 2-folds) 20 vertices, 12 faces, 30 edges (10 3-folds, 6 5-folds, 15 2-folds) Asymmetric unit is 1/60th of whole object Object consists of 60 identical subunits arranged with icosahedral symmetry NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

16 Icosahedral (532) Point Group Symmetry 30 dimers 20 trimers 12 pentamers From Eisenberg & Crothers, Table 16-3, p.767

17 3D Reconstruction of Icosahedral Particles Symmetry 1. Icosahedral (532) point group symmetry 2. Triangulation symmetry Purely mathematical concept (concerns lattices) Real objects (e.g. viruses) with 532 symmetry often consist of multiples of 60 subunits Subunits arranged such that additional, local or pseudosymmetries exist NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

18 T=13l T=13laevo T=16 Reovirus Rotavirus HSV T=4 T= 2 T=7d T=7 T=7dextro NωV LA-1 Polyoma CaMV HPV T=4 Ross River T=4 T=1 T=1 T=3 T=1 T=1 T=1 T=3 T=1 NβV SpV-4 φx174 FHV HRV-14 Polio CPMV 500 Å CCMV B19

19 T=16 T=169d HSV T=4 T= 2 T=7d NωV LA-1 Polyoma PBCV-1 T=4 T=1 T=1 T=3 T=1 T=1 T=1 T=3 T=1 NβV SpV-4 φx174 FHV HRV-14 Polio CPMV 500 Å CCMV B19

20 3D Reconstruction of Icosahedral Particles Triangulation Number Key Concept: T symmetry is NOT incorporated into or enforced by the 3D reconstruction algorithms Hence, T symmetry emerges as a result of a properly performed 3D reconstruction analysis NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

21 3D Reconstruction of Icosahedral Particles Triangulation Number Key Concept: T symmetry is NOT incorporated into or enforced by the 3D reconstruction algorithms In other words: What you determine is the structure of one asymmetric unit of the object NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

22 3D Reconstruction of Icosahedral Particles Two Basic Assumptions: - Specimen consists of stable particles with identical structures (else averaging is invalid) - Programs test for and assume presence of icosahedral (532) symmetry NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

24 3D Reconstruction of Icosahedral Particles Protocol Electron Cryo-Microscopy Sample : ~2-3 µl at 1-5 mg/ml Specimen support: holey carbon film (1-2 µm) 1 µm

25 3D Reconstruction of Icosahedral Particles Protocol Electron Cryo-Microscopy Sample : ~2-3 µl at 1-5 mg/ml Specimen support: holey carbon film (1-2 µm) 1 µm

26 3D Reconstruction of Icosahedral Particles Protocol Electron Cryo-Microscopy Sample : ~2-3 µl at 1-5 mg/ml Specimen support: holey carbon film (1-2 µm) 100 nm

27 3D Reconstruction of Icosahedral Particles Protocol Electron Cryo-Microscopy Sample : ~2-3 µl at 1-5 mg/ml Specimen support: holey carbon film (1-2 µm) Microscope: kev with FEG I think I may have spotted something!

28 3D Reconstruction of Icosahedral Particles Protocol Electron Cryo-Microscopy Sample : ~2-3 µl at 1-5 mg/ml Specimen support: holey carbon film (1-2 µm) Microscope: kev with FEG Defocus range: 1-3 µm underfocus Dose: e-/å2 Film: SO-163 (12 min, full strength) Micrographs: >1000s(?) Particles: > > 106 (?) Target resolution: 10-6 Å --> 4Å (?) FEI Tecnai F30 Polara

29 100 nm

31 Icosahedral Particle Image Reconstruction Scheme? Black Box

32 Icosahedral Particle Image Reconstruction Scheme Digitize Micrograph Graphics Box G y x Determine Ιντερπρετ Ρεφινε Ιντερπαρτιχλε ετεχτ Νορµαλιζε θ Ξ Ψ Ζ φ ω Β ξψ Παρτιχλε Χοµπυτε αχκ Πυβλιση et ισπλαψ/ινσπεχτ/αναλψζε Ρ Ρεγυλαρ εφινε initial µορε out προϕεχτ παρτιχλε ιµαγε θ,φ and παρτιχλ θ,φ,ω,ξ,ψ στρυχτυρε 3 estimate ορ ιµαγε Ρεφινε ϖαλυεσ αρτιφαχτσ γριδ ρεχονστρυχτιον float περιση 3 οριεντατιον εοριγινσ µεανσ/ϖαριανχεσ στρυχτυρε οφ ιµαγεσ οριγινσ παραµετερσ θ,φ of (αστιγµατισµ, ρεφινεµεντ Ιχοσαηεδραλ ασψµµετριχ υνιτ:?αδδ ΞΨΖ ϑουρναλ ϖιρυσ χαυσεσοφ µυχη παιν ανδ συφφερινγ ιν µεν, ςιρυσ ωοµεν, ανδστρυχτυρε χηιλδρεν... Ινχρεασε micrograph display individual particle Χορρελατε δριφτ, Ρεµοϖε Ιµποσε (Χροσσ χορρελατε ετχ.) βαχκγρουνδ 3 centers ρεσολυτιον particle ανδ στρυχτυρε ωιτη 532 σψµµετρψ εστιµατε οτηερ (x,y γραδιεντσ (ε.γ. ιµαγεσ ρεσολυτιον βιοχηεµιχαλ, ωιτη ιν ετχ.) δατα οριεντατιον (Χοµπαρε ϖαλυεσ ιν ιµαγεσ ασψµµετριχ ωιτη προϕεχτεδ υνιτ Σηαδεδ τριανγλε βουνδεδ βψ 2 φολδ, images origins) Φουριερ προϕεχτεδ τρανσφορµσ ϖιεωσ οφ 3 ρεχονστρυχτιον) ανγλεσ ϖιεωσ (θ,φ,ω) οφ 3 ρεχονστρυχτιον) 3 φολδ, ανδ τωο αδϕαχεντ 5 φολδ αξεσ

33 Icosahedral Particle Image Reconstruction Scheme Black Box

34 Expand Data Set Y N Digitize Micrograph PSC, PhotoScan Uniconv RobEM Box Particles RobEM Pre-Process Images RobEM AutoPP Establish Refinement Criteria: Resolution Limits Refinement More Data? Resample Rebox PFTSEARCH, OOR, PO2R Select Images Select Compute 3DR EM3DR, P3DR Monitor Data Quality Y Refine? Initial 3D Model Determine Origin and Orientation (θ,φ,ω,x,y) Refine CTF N PROGRAM(S) N RobEM, PCUT, PSF CUTSHELL MAPFFT, EMRES Done? Y Visualize & Interpret 3D Structure Archive Data RobEM, O, Chimera, Explorer, EMfit, Situs, PIF2ASC, Uniconv MAPMAN Etc.

35 Expand Data Set Digitize Micrograph Box Particles Y N Pre-Process Images Establish Refinement Criteria: Resolution Limits Refinement More Data? Resample Rebox Determine Origin and Orientation (θ,φ,ω,x,y) PFTSEARCH, OOR, PO2R Select Images Select Compute 3DR EM3DR, P3DR Refine CTF Monitor Data Quality Y N Refine? Initial 3D Model N Done? Y Visualize & Interpret 3D Structure Archive Data RobEM, PCUT, PSF CUTSHELL MAPFFT, EMRES

36 Expand Data Set Digitize Micrograph Box Particles Y N Pre-Process Images Establish Refinement Criteria: Resolution Limits Refinement More Data? Resample Rebox Determine Origin and Orientation (θ,φ,ω,x,y) Refine CTF Select Images Compute 3DR Monitor Data Quality Y N Initial 3D Model Refine? N Done? Y Visualize & Interpret 3D Structure Archive Data AUTO3DEM

37 Expand Data Set Digitize Micrograph Box Particles Y N Pre-Process Images Establish Refinement Criteria: Resolution Limits Refinement More Data? Resample Rebox Determine Origin and Orientation (θ,φ,ω,x,y) Refine CTF Select Images Compute 3DR Monitor Data Quality Y N Refine? N Done? Y Visualize & Interpret 3D Structure Archive Data Initial 3D Model

38 Digitize Micrograph

39 Digitize Micrograph Box Particles

40 Digitize Micrograph Box Particles Extracted

41 Digitize Micrograph Box Particles Masked

42 Digitize Micrograph Box Particles Floated

43 Digitize Micrograph Box Particles Apodized

44 Digitize Micrograph Box Particles Square mask; unfloated

45 Digitize Micrograph Box Particles Circular mask; unfloated

46 Digitize Micrograph Box Particles Circular mask; floated

47 Digitize Micrograph Box Particles Circular mask; floated & apodized

48 Digitize Micrograph Box Particles

49 Digitize Micrograph Box Particles Pre-Process Images Remove blemish, Remove Gradient Normalize means/variances, Apodize Determine CTF parameters Create Initial Parameter Files

50 Pre-Process Images Remove blemish, Remove Gradient Normalize means/variances, Apodize Determine CTF parameters Create Initial Parameter Files

52 Pre-Process Images Remove blemish, Remove Gradient Normalize means/variances, Apodize Determine CTF parameters Create Initial Parameter Files Gradient removed

53 Pre-Process Images Remove blemish, Remove Gradient Normalize means/variances, Apodize Determine CTF parameters Create Initial Parameter Files Gradient not removed

58 Pre-Process Images Remove blemish, Remove Gradient Normalize means/variances, Apodize Determine CTF parameters Create Initial Parameter Files Extracted Masked Floated Apodized

67 Digitize Micrograph Box Particles Pre-Process Images Establish Refinement Criteria: Resolution Limits

68 Digitize Micrograph Box Particles Pre-Process Images Establish Refinement Criteria: Resolution Limits Determine Origin and Orientation (θ,φ,ω,x,y)

69 Determine Origin and Orientation (θ,φ,ω,x,y) Goal: determine phase origin and view orientation for each boxed particle MOST IMPORTANT STEP? Garbage in > garbage out

70 Determine Origin and Orientation (θ,φ,ω,x,y) People who don t know which end is up

71 Specifying Direction of View: (θ,φ,ω) Orientation Z (0,0) Standard Setting θ 3 5 ω φ (80,-15) X (90,0) 5 Y (90,90)

72 BPV Projections: Icosahedral ASU Z 3 5 θ θ 3 φ X 5 Y 5 2 φ 5

73 BPV Projections: 1/2 Icosahedral ASU θ φ

74 Initial 3D Model Determine Origin and Orientation (θ,φ,ω,x,y) How do we determine the (θ, φ, ω, x, y) parameters? Two methods: 1. Common lines New or unknown structure 2. Model-based (template) matching General features of structure are known or a crude model can be generated

75 Determine Origin and Orientation (θ,φ,ω,x,y) How do we determine the (θ, φ, ω, x, y) parameters? Two methods: 1. Ab initio (e.g. Common lines) New or unknown structure 2. By guess and by golly

76 Initial 3D Model Determine Origin and Orientation (θ,φ,ω,x,y) How do we determine the (θ, φ, ω, x, y) parameters? Two methods: 1. Ab initio (e.g. Common lines) New or unknown structure 2. Model-based (template) matching General features of structure are known or a crude model can be generated ( or, sometimes, even a lousy model will work)

77 Determine Origin and Orientation (θ,φ,ω,x,y) Common Lines The gospel according to Tony Crowther (Phil. Trans. R. Soc. Lond. B.(1971) 261: ) [Common lines] arise as follows: An observed section of the transform intersects an identical symmetry-related section in a line, along which the transform must have the same value in both sections The common line lies in the original section. However, regarded as lying in the symmetry-related section it must have been generated by the symmetry operation from some other line in the original section.

78 Determine Origin and Orientation (θ,φ,ω,x,y) Common Lines The gospel continued: We therefore have a pair of lines in the original transform plane along which the transform must have identical values A similar pair of lines will be generated by each possible choice of pairs of symmetry operations The angular positions of these lines are dependent on the orientation of the particle.

79 Orientation Determination by Common Lines 3D Object 2D Projection (θ,φ,ω) 2D Fourier Transform

80 Orientation Determination by Common Lines Simple example: object with single three-fold axis of symmetry X B 1 C ABCD = 2D transform of image from particle not viewed along an axis of symmetry A A 2 Z Y Let Z-direction coincide with 3-fold axis of symmetry D 3-fold operation generates two additional FT sections (only A B C D is shown) D 3 B C Adapted from Moody (1990) Fig. 7.68, Both planes have common values along the line (1,2,3) of their intersection

81 Orientation Determination by Common Lines X A B 1 C B A B 1 1 A 2 2 A Z Y D C D C AA BB 1 D 3 B 1 2 C 3 CC Adapted from Moody (1990) Fig. 7.68, D 3 DD Adapted from Moody (1990) Fig. 7.69, p.246

82 Orientation Determination by Common Lines Original Transform Plane

83 Orientation Determination by Common Lines Symmetry-Related Transform Plane 3 3

84 Orientation Determination by Common Lines Ok, that s easy (simple object with single 3-fold axis) What about an object with 532 symmetry? For a general view, icosahedral symmetry generates: 5-folds: 12 x 2 = 12 pairs 2 3-folds: 20 x 1 = 10 pairs 2 2-folds: 30 x 1 = 15 real lines 2 37 common lines

85 Orientation Determination by Common Lines (80,11) (69,0) (89,-1)

86 Orientation Determination by Common Lines (80,11)

87 Orientation Determination by Common Lines (80,11) What is (θ,φ,ω) for this particle?

88 Orientation Determination by Common Lines ω (80,11,0)

92 1 1 2 Orientation Determination by Common Lines 2 ω (80,11,90)

93 Orientation Determination by Common Lines 1 2 ω 2 (80,11,180) 1

94 Orientation Determination by Common Lines (80,11,ω) 1 Metric: Identify ω that gives lowest phase residual

95 Orientation Determination by Common Lines Repeat process for all possible (θ,φ,ω) combinations

96 Orientation Determination by Common Lines > 250,000 combinations for 1 angular search intervals

97 Determine Origin and Orientation (θ,φ,ω,x,y) Common Lines The (θ, φ, ω) that results in the lowest phase residual is selected as the best estimate for the particle view orientation The common lines procedure is similarly used to determine the particle phase origin (x, y) Not to worry.. I ll spare you the details!!!

98 Initial 3D Model Determine Origin and Orientation (θ,φ,ω,x,y) Recall: two methods to determine (θ, φ, ω, x, y): 1. Common lines 2. Model-based (template) matching Bulk of structures now solved this way Details discussed in practical session

99 PFTSEARCH Program Flowchart Project Model (m views) Interpolate Projections m Projections 3D Model Polar PRJs (θ,φ)j Projection (θ,φ,0)j Select best images Compute 3DR Rotate by ωj Sum average all projections (θ,φ,ω,,) ωj Projection (θ,φ,ω)j j=1,n Cross-correlate --> (x,y)j Circular Average j j n Images (j=1,n) ( 0, ) 0 j m Polar FTs (θ, φ )j Polar PRJ (θ, φ,0)j Invert γ Polar PRJ (θ, φ,0)j (θ,φ,ω)j (,) Cross-correlate --> (xo,yo)j m 1-D FFT Rotationally correlate --> (φ,ω)j (θ, φ )j Correlate --> (θ, φ )j j Interpolat e Imagej 1-D FFT Polar imagej Polar FTj

100 Initial 3D Model Determine Origin and Orientation (θ,φ,ω,x,y) ID θ φ ω x y

101 Digitize Micrograph Box Particles Pre-Process Images Establish Refinement Criteria: Resolution Limits Determine Origin and Orientation (θ,φ,ω,x,y) Initial 3D Model

102 Digitize Micrograph Box Particles Pre-Process Images Establish Refinement Criteria: Resolution Limits Determine Origin and Orientation (θ,φ,ω,x,y) Select Images Initial 3D Model

103 Select Images Goal: weed out bad particle images before computing 3D reconstruction

104 Select Images ID θ φ ω x y

105 Select Images PRJ CC PFT CC CMP CC ? ?

106 Select Images PFT Coefficient ± 0.075

107 Digitize Micrograph Box Particles Pre-Process Images Establish Refinement Criteria: Resolution Limits Determine Origin and Orientation (θ,φ,ω,x,y) Select Images Initial 3D Model

108 Digitize Micrograph Box Particles Pre-Process Images Establish Refinement Criteria: Resolution Limits Determine Origin and Orientation (θ,φ,ω,x,y) Select Images Compute 3DR Initial 3D Model

109 Compute 3DR Goal: combine only good particle images to compute a 3D density map

110 Compute 3DR In theory In practice From Lake (1972), p.174

111 Compute 3DR g ρ G Overall scheme: F

112 Compute 3DR Steps: Z 1. Compute 2D FFT of each particle image Φ 2. Combine all 2D FFTs to build up 3D Fourier-Bessel transform R Central section F(R,Φ,Z) Crowther, DeRosier and Klug, 1970, p.329 Adapted from Crowther (1971) Fig. 4, p.223

113 g ρ G Compute 3DR F Steps: 1. Compute 2D FFT of each particle image 2. Combine all 2D FFTs to build up 3D Fourier-Bessel transform Compute Gn s on each annulus G = ( B B) B F 4. Compute gn s from Gn s (Fourier-Bessel transform) 5. Compute polar density map (ρ(r,φ,z)) from gn s 6. Convert from polar to Cartesian map ( ρ(r,φ,z) --> ρ(x,y,z) )

114 Digitize Micrograph Box Particles Pre-Process Images Establish Refinement Criteria: Resolution Limits Initial 3D Model Determine Origin and Orientation (θ,φ,ω,x,y) Select Images Compute 3DR Option: correct for CTF effects in particle FFTs before FFTs are merged to form the 3D FFT

115 Digitize Micrograph Box Particles Pre-Process Images Establish Refinement Criteria: Resolution Limits Determine Origin and Orientation (θ,φ,ω,x,y) Select Images Compute 3DR Monitor Data Quality Initial 3D Model

116 Monitor Data Quality Goal: assess resolution of 3D density map to determine what to do next

117 Select Images Compute 3DR Monitor Data Quality

118 Select Images Even Images Odd Images Compute Even 3DR Compute Odd 3DR Mask Even 3DR (r1-r2) Mask Odd 3DR (r1-r2) Unmasked Reovirus

119 Select Images Even Images Odd Images Compute Even 3DR Compute Odd 3DR Mask Even 3DR (r1-r2) Mask Odd 3DR (r1-r2) Masked Reovirus

120 Select Images Even Images Odd Images Compute Even 3DR Compute Odd 3DR Mask Even 3DR (r1-r2) Mask Odd 3DR (r1-r2) FT Even SFs FT Odd SFs

121 Select Images Even Images Odd Images Compute Even 3DR Compute Odd 3DR Mask Even 3DR (r1-r2) Mask Odd 3DR (r1-r2) FT Even SFs FT Odd SFs Compute and plot phase differences and correlation coefficients as function of spatial frequency

122 Correlation Coefficient Monitor Data Quality Spatial Frequency (1/Å) 8.7Å 6.9Å Unmasked Reovirus Masked Reovirus

123 Monitor Data Quality Note: quality of 3D density map is not identical everywhere in the map

124 Resolution Monitor Data Quality 7Å Reovirus Radius r1 = 245 r2 = 363

127 Monitor Data Quality

134 Digitize Micrograph Box Particles Pre-Process Images Establish Refinement Criteria: Resolution Limits Determine Origin and Orientation (θ,φ,ω,x,y) Select Images Compute 3DR Monitor Data Quality Initial 3D Model

135 Digitize Micrograph Box Particles Pre-Process Images Establish Refinement Criteria: Resolution Limits Determine Origin and Orientation (θ,φ,ω,x,y) Select Images Compute 3DR Monitor Data Quality Done? Initial 3D Model

136 Digitize Micrograph Box Particles Pre-Process Images Establish Refinement Criteria: Resolution Limits Determine Origin and Orientation (θ,φ,ω,x,y) Select Images Compute 3DR Monitor Data Quality Refine? N Done? Initial 3D Model

137 Digitize Micrograph Box Particles Pre-Process Images Refinement Establish Refinement Criteria: Resolution Limits Determine Origin and Orientation (θ,φ,ω,x,y) Refine CTF Select Images Compute 3DR Monitor Data Quality Y Refine? N Done? Initial 3D Model

138 Digitize Micrograph Box Particles More Data? Pre-Process Images Refinement Establish Refinement Criteria: Resolution Limits Determine Origin and Orientation (θ,φ,ω,x,y) Refine CTF Select Images Compute 3DR Monitor Data Quality Y N Refine? N Done?

139 Expand Data Set Digitize Micrograph Box Particles Y N Pre-Process Images Establish Refinement Criteria: Resolution Limits Refinement More Data? Resample Rebox Determine Origin and Orientation (θ,φ,ω,x,y) Refine CTF Select Images Compute 3DR Monitor Data Quality Y N Refine? N Done?

140 Expand Data Set Digitize Micrograph Box Particles Y N Pre-Process Images Establish Refinement Criteria: Resolution Limits Refinement More Data? Resample Rebox Determine Origin and Orientation (θ,φ,ω,x,y) Refine CTF Select Images Compute 3DR Monitor Data Quality Y N Refine? N Done? Y Visualize & Interpret 3D Structure

141 Expand Data Set Digitize Micrograph Box Particles Y N Pre-Process Images Establish Refinement Criteria: Resolution Limits Refinement More Data? Resample Rebox Determine Origin and Orientation (θ,φ,ω,x,y) Refine CTF Select Images Compute 3DR Monitor Data Quality Y N Refine? N Done? Y Visualize & Interpret 3D Structure Archive Data

143 3D Reconstruction of Icosahedral Particles Outline - Background - References; examples; etc. - Symmetry - Icosahedral (532) point group symmetry - Triangulation symmetry - Typical procedure (flow chart) - Digitization and boxing - Image preprocessing / CTF estimation - Initial particle orientation/origin search - Orientation/origin refinement - 3D reconstruction with CTF corrections - Validation (resolution assessment) - Current and future strategies NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

144 3D Reconstruction of Icosahedral Particles Current and Future Strategies - Parallelization and new algorithms - Parallel versions of EM3DR, PFTSEARCH, OOR - EM3DR ---> P3DR - OOR > PO2R - Automation - Semi-auto boxing (RobEM) - Automated origin/orientation refinement (AUTO3DEM) - Split data set processing - Divide image data at very beginning and refine even and odd data independently. - Minimizes (eliminates?) bias in resolution assessment - Combine independent reconstructions to obtain final 3DR with highest S/N NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

145 Structure Determination Flow Chart Initial model A Boxed particles Initial model B Search (θ,φ,ω,x,y) Split Search (θ,φ,ω,x,y) Select particles Search Select particles Compute new model A Compute new model B Check resolution N Search Done? N Y Refine Refine (θ,φ,ω,x,y) Refine (θ,φ,ω,x,y) Select particles Select particles Compute new model A Compute new model B Check resolution N Refine Done? N Y Compute final map NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

146 Digitize micrographs Data Flow for Split Data Set Processing Box particles Pre-process images odd even Initial 3D Model Orientation Search Orientation Search Select Select Compute odd map Compute even map Check resolution N Done searching? SEARCH Initial 3D Model N Refine (θ,φ,ω,x,y) NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)

147 3D Reconstruction of Icosahedral Particles Current and Future Strategies - Data selection Trying to improve resolution by substantially increasing the number of images averaged ad infinitum may prove less beneficial than simply applying more rigorous quality control measures to weed out bad data. Borgnia, M. J., D. Shi, P. Zhang and J. L. Milne (2004) Visualization of α-helical features in a density map constructed using 9 molecular images of the 1.8 MDa icosahedral core of pyruvate dehydrogenase. J. Struct. Biol. 147: Worst 138 of 963 NRAMM Workshop, La Jolla, CA (Nov 2-10, 2005)