Space group validation using Zanuda Andrey Lebedev, CCP4
Slide 2
09/01/2015CCP4 Study Weekend Pseudosymmetry 2 Crystallographic
symmetry Global and exact General non-crystallographic symmetry
Local and not exact Pseudo-symmetry Global and not exact Global
consistent with crystallographic translations pseudosymmetry
space/point groups can be defined
Slide 3
09/01/2015CCP4 Study Weekend Pseudosymmetry and twinning 3 twin
axis Twinning + Generic NCS Twinning axis || NCS axis Single
crystal OR
Slide 4
Pointless: twinning and pseudo-symmetry 09/01/2015CCP4 Study
Weekend4 Nelmt...CC...Rmeas... operator... 1...0.98...0.051identity
2...0.98...0.071*** 2-fold l... 3...0.96...0.089*** 2-fold h...
4...0.97...0.089*** 2-fold k... Laue Group...CC...Rmeas... = 1 P m
m m ***...0.97...0.07... 2 P 1 2/m 1...0.98...0.06... 3 P 1 2/m
1...0.97...0.07... 4 P 1 2/m 1...0.97...0.07... 5 P
-1...0.98...0.05... "The second best" point group may be
correct
Slide 5
09/01/2015CCP4 Study Weekend Pseudotranslation 5
Crystallographic translation: global and exact Pseudo-translation:
global but not exact Pseudotranslation + crystallographic axes
-> pseudosymmetry axes
Slide 6
Pseudotranslation: what else can go wrong? 09/01/2015CCP4 Study
Weekend6 Molecular Replacement: Two structures are globally very
similar (e.g. rmsd = 0.5A) MR can in some cases pick up a wrong
solution Crystallographic and pseudosymmetry axes are confused P2
(a,b,c) Cell and H-M symbol are the same True structureFalse
structure
Slide 7
True structureFalse structure Example: anti-TRAP 09/01/2015CCP4
Study Weekend7
Slide 8
True structure: R26% R-free32% complete model can be built
False structure: R38% R-free44% partial model only can be built
True and false structures: refinement and rebuilding (P21 true and
P21 false structures of anti-TRAP from B. licheniformis)
09/01/2015CCP4 Study Weekend8
Slide 9
http://www.ccp4.ac.uk/ccp4online Zanuda CCP4 online
09/01/2015CCP4 Study Weekend9
Slide 10
mtz-file pdb-file Submitting Zanuda job 09/01/201510CCP4 Study
Weekend
Slide 11
Download of output pdb- and mtz-files Symmetry analysis Zanuda
output 09/01/201511CCP4 Study Weekend
Slide 12
False space groupP2 P2 1 True space group C2Pseudo-symmetry
space group Space group and its relation to the structure 1yup
Crystallographic axes Pseudosymmetry axes Positions of molecules A
simple example (1yup) 09/01/201512CCP4 Study Weekend
Slide 13
Space group validation: step 1 09/01/2015CCP4 Study Weekend13
Pseudosymmetry Space Group; here: a' = a/2
Slide 14
Space group validation: step 2 09/01/2015CCP4 Study Weekend14
2-axis2 1 -axis C2crystallographic crystallographic
P2crystallographic NCS P2 1 NCS crystallographic True SG
Slide 15
space group validation: step 3 09/01/2015CCP4 Study Weekend15
Output (P2 1 )
Slide 16
Example from prof Daan Van Aalten, Dundee Vicious cycle: good
model is needed to clearly distinguish between correct and
incorrect space groups knowledge of correct space group is needed
(desirable) to improve model Availability of P1 MR solution can be
beneficial data reduction in zanuda is not perfect reprocess data
with aimless, if space group has to be changed (aimless job 88;
zanuda jobs 17, 82; refmac jobs 74, 90) 09/01/2015CCP4 Study
Weekend16
Slide 17
Example from Misha Isupov, Exeter Four alternative origins
(alternative P21) Even good programs can make mistakes, and even in
simple cases: solved with Phaser corrected with Zanuda Simple
expansion of an incorrect model into P1 and refinement may not work
(China54: phaser job 1; refmac jobs 2; zanuda job 7; coot on any
structure) 09/01/2015CCP4 Study Weekend17
Slide 18
CCP4I interface 09/01/2015CCP4 Study Weekend18 "REFINE" or
"SAVE" only (quick) Do not "SYMMETRISE" for MR solutions in
P1!
Slide 19
09/01/2015CCP4 Study Weekend19
Slide 20
09/01/2015CCP4 Study Weekend20
Slide 21
09/01/2015CCP4 Study Weekend21
Slide 22
mosflm pointless aimless ctruncate structure solution model
building and refinement PDB imosflm ccp4 automatic space group
assignment: - analysis of lattice symmetry - comparison of related
reflections - analysis of systematic absences Automatic space group
assignment 09/01/2015CCP4 Study Weekend22
Slide 23
mosflm pointless aimless ctruncate structure solution PDB
imosflm ccp4 pointless aimless ctruncate structure solution PDB
twin not solved try second best space group Manual space group
assignment 09/01/2015CCP4 Study Weekend23 model building and
refinement model building and refinement
Slide 24
mosflm pointless aimless ctruncate structure solution PDB
imosflm ccp4 R-free 40% density does not improve Zanuda Attempts at
refinements in all relevant subgroups of pseudo-symmetry space
group Strong pseudo-symmetry 09/01/2015CCP4 Study Weekend24 model
building and refinement some model rebuilding reiterate data
reduction etc.
Slide 25
Pseudosymmetry: what can go wrong 09/01/2015CCP4 Study
Weekend25 Data reduction: Pseudosymmetry can be confused with the
crystallographic symmetry Wrong space group assignment Problems
with structure solution Most relevant cases Twin axis ||
pseudosymmetry axis Not unusual
Slide 26
PDB code:1yup space group (PDB):P18 molecules per a.u. space
group (true):P2 1 4 molecules per a.u. Pseudo-symmetry space
group:C22 molecules per a.u. (because of pseudo-translation) An
example of symmetry correction 09/01/201526CCP4 Study Weekend
Zanuda protocol is not perfect Assumptions: The pseudosymmetry
is very strong (r.m.s.d. from exact symmetry 1A) The structure is
almost correct (although it might have been refined / rebuilt in an
incorrect space group) If it is not so, then it is unlikely to
obtain the correct answer. 09/01/2015CCP4 Study Weekend28 Things
went wrong way
Slide 29
Summary Signs of incorrect interpretation of pseudo-symmetry:
Reasonable density for a part of the structure Large fragments of
poor density with no useful features for further rebuilding High
R-free However, model building and refinement in incorrect space
group may improve model Then models with different symmetries can
be generated, refined and compared 09/01/2015CCP4 Study
Weekend29
Slide 30
Zanuda: space group validation Algorithm: From input
model:determine pseudosymmetry space group (PSSG) From PSSG:select
subgroups with observed unit cell For each such subgroup: Convert
model and data into the subgroup Restrained refinement Repeat
refinements of the best (R-free) model Starting from P1 Adding the
best (r.m.s.d.) symmetry element at each refinement Terminate if
there is no symmetry elements to be added Terminate and cancel the
last symmetry element if R-free jumps 09/01/2015CCP4 Study
Weekend30
Slide 31
Zanuda: limitations Assumptions: The pseudosymmetry is very
strong (r.m.s.d. from exact symmetry 1A) The structures of
individual molecules are almost correct although they might have
been refined / rebuilt in an incorrect space group If assumptions
are not satisfied, the results will likely to be wrong.
09/01/2015CCP4 Study Weekend31
Slide 32
Model preparation; determination of pseudosymmetry space group
including pseudotranslation Subgroup/supergroup graph Test
refinements in subgroups with observed unit cell (blue) Best
structure expanded into P1 Refinements followed by adding symmetry
elements Returns model and data in the "best" SG Algorithm Common
supergroup: P4 2 22 with twice smaller c 09/01/201532CCP4 Study
Weekend
Slide 33
Options: - currently available:refinement in all possibilities
SGs (auto) - to be added:refinements in subgroups selected by a
user - to be added:transformations only, no refinements -
Appropriate interface - Will be OK for distribution Using unmerged
data as an option - this is the right way - using Pointless for
data conversion Starting from incomplete model (completion by e.g.
Molecular Replacement) 09/01/201533CCP4 Study Weekend Possible
development
Slide 34
09/01/2015CCP4 Study Weekend Four alternative solutions in two
space groups 34 GAF (N-terminal) domain of CodY protein from
Bacillus subtilis Levdikov, V. M. et al. (2006). J Biol Chem 281,
11366-73. MR solution
Slide 35
09/01/2015CCP4 Study Weekend Comparison of possible structures
35
Slide 36
Twinned crystal with pseudo-symmetric substructure
09/01/2015CCP4 Study Weekend36 Substructure (A) is common for twin
individuals Substructure (B) is not even approximately symmetric
relative to and The choice of correct origin was essential for
structure completion (A)(B) + 3-fold axes with respect to the true
structure: crystallographic pseudosymmetry for (A) pseudo-P3 1 21
(a' b' c)P3 1 (a b c) Human macrophage receptor CLEC5A for dengue
virus Watson, A. A. et al. (2011). J Biol Chem 286, 24208-18.
Slide 37
Attempt at structure solution 09/01/2015CCP4 Study Weekend37
Conventional MR (4 monomers) Phased MR (2 monomers) Refinement R =
0.45 R-free = 0.48 Something is wrong False origin?
Slide 38
Pseudo-symmetry space group 09/01/2015CCP4 Study Weekend38
Hermann Mauguin symbol P3 1 21 translation base vectors b' = (a -
b)/ 3 a' = (a + 2b)/ 3 c' = c in the true large cell three
alternative origins associated with the three types of 3-fold axes
shown in the figure 3-fold axes with respect to MR solution:
crystallographic pseudosymmetry
Slide 39
(A) (C) Transformation of MR solution 09/01/2015CCP4 Study
Weekend39 We need to move structure to tell the program that we
want to select different set of axes Select the AU, in which 3
pairs of molecules are related by pseudotranslation save required
symmetry equivalents coot merge them into a single PDB-file text
editor Shift the whole AU by (a + b)/ 3 lsqkab structurereference
(unit cell) This picturefixedmoves Refinement program
movesfixed
Slide 40
Using Zanuda 09/01/2015CCP4 Study Weekend40 Table form Zanuda:
comparison of refinements R-factors are still too high Incomplete
structure? (C) P10.4300.466 P3 1 0.4600.498 P3 1 0.4590.495 P3 1
0.4300.466 C20.4410.481 P3 1 120.4550.480
Slide 41
Crystal disorder Twinning, partial disorder:Missing global
periodicity 09/01/2015CCP4 Study Weekend41 Single crystal(Single
ordered domain) Twinned crystal(Two or more ordered domains)
Partially disordered crystal(Many ordered domains) size of ordered
domains Coherence length of X-rays
Slide 42
09/01/2015CCP4 Study Weekend42
Slide 43
09/01/2015CCP4 Study Weekend43
Slide 44
09/01/2015CCP4 Study Weekend44
Slide 45
Guidelines In most cases high R-free means wrong solution or
insufficiently good model Zanuda can also make mistakes especially
with poor starting model Hence: Firstly, improve model Secondly,
improve model If nothing works, then perhaps space group or origin
assignment is a problem Model from MR: R-free 50-60%: first of all,
this structure needs to be solved (or model rebuilt) R-free around
40% after model rebuilding and no obvious ways for further
improvement of the model: try Zanuda 09/01/2015CCP4 Study
Weekend45
Slide 46
Zanuda 09/01/2015CCP4 Study Weekend46 CCP4I default & YSBL
server - Automatic detection of pseudosymmetry space group -
Refinement in subgroups of PSSG - Selection and validation of the
best model CCP4I option - no refinement, save all transformed
models and data When can be used: (1) Space group validation for
twinning || pseudosymmetry (2) False origin correction (as in the
CodY example) (3) Structure solver in P1
Slide 47
Pseudotranslation 09/01/2015CCP4 Study Weekend47
Crystallographic translation Pseudo-translation Pseudotranslation
C/2 Planes 2L+1 contain weak reflections Limiting case, C' = C/2
Weak reflections vanish Two times larger reciprocal lattice spacing
Crystallographic translation
Slide 48
> Acentric moments of E for k=1,3,4 > 4th moments of E...
1jjk: Pseudotranslation results in alteration of strong and weak
reflections Statistics of one intensity are strongly affected by
pseudotranslation 09/01/201548CCP4 Study Weekend
Slide 49
False structure P2P2 1 True structure C2 Pseudo-symmetry space
group Crystallographic axes Pseudosymmetry axes Monoclinic groups
related to 1yup 09/01/2015CCP4 Study Weekend49
Slide 50
CCP4I interface CCP4I > Validation & Deposition >
Validate space group Modes REFINE all transformed models and save
the best model SAVE all transformed models and data without
refinement Option SYMMETRYSE input model before further refinements
09/01/2015CCP4 Study Weekend50