Re: [ccp4bb] DM NCS averaging question
Well - if the CCs are 0 then no averaging can take place.
You know you can let DM make the mask itself - are you using the GUI?
It shows you what to set..
Eleanr
zhan...@umbc.edu wrote:
Hi,
Thanks for reminding me checking the mask. I think their might be
something wrong with the mask, since when DM read in the mask, it says:
Number of columns, rows, sections ... 84 74 69
Map mode 0
Start and stop points on columns, rows, sections -53 30
80 153 -4 64
Grid sampling on x, y, z 136 260 150
Cell dimensions . 135.57100
260.11200 150.2 90.0 101.14000 90.0
Fast, medium, slow axes .ZXY
Minimum density . 0.0
Maximum density . 0.0
Mean density 0.0
Rms deviation from mean density . 0.0
Space-group .4
Number of titles 1
It seems the mask is just null. However, I converted it to a map file, and
coot clearly showed the mask, so I am not sure why the null mask was found
by DM. Moreover, the NCS CCs are just 0s for the mask.
Anyway, following is my NCSMASK script I used to generate the above mask,
where XYZIN is the reorganized pdb containing only a single fixed NCS unit
(chain A). and all the operations were generated by LSQKAB with Chain A
mapped to other chains. Not sure whether there is something wrong here or
not...
ncsmask xyzin ${PDB}_A.pdb mskout ${PDB}.msk eof
SYMM P1211
EXPAND 1.0
OVERLAP 3
AVERAGE 12
#Identical
ROTA MATRIX 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
TRAN 0.0 0.0 0.0
#AC
ROTA MATRIX -0.22748 0.97259 0.04813 -0.97372 -0.22662 -0.02273
-0.01120 -0.05203 0.99858
TRAN 101.4683781.74413 2.89341
#AD
ROTA MATRIX -0.90337 0.42792 0.02831 -0.42883 -0.90066 -0.07011
-0.00451 -0.07547 0.99714
TRAN 158.0331736.91842 3.25853
#AF
ROTA MATRIX -0.21272 -0.97702 0.01352 0.97675 -0.21300 -0.02424 0.02657
0.00805 0.99961
TRAN 100.69797 -81.01860-1.71365
#AG
ROTA MATRIX 0.61704 -0.78687 -0.01005 0.78590 0.61553 0.05899 -0.04023
-0.04429 0.99821
TRAN 32.50667 -65.76570 7.05504
#AH
ROTA MATRIX -0.85814 -0.51116 -0.04813 -0.51290 0.85771 0.03558
0.02310 0.05522 -0.99821
TRAN 156.1498142.2387348.93406
#AI
ROTA MATRIX -0.13703 -0.98975 -0.04032 -0.99048 0.13635 0.01902
-0.01332 0.04254 -0.99901
TRAN 95.9763082.3094852.82510
#AJ
ROTA MATRIX 0.69662 -0.71695 -0.02645 -0.71716 -0.69691 0.00230
-0.02008 0.01737 -0.99965
TRAN 25.8458859.7628653.68224
#AK
ROTA MATRIX 0.99467 0.10258 -0.01072 0.10259 -0.99472 0.00088 -0.01057
-0.00197 -0.4
TRAN0.47215-8.8608252.78315
#AL
ROTA MATRIX 0.55782 0.82946 0.02875 0.82987 -0.55793 -0.00466 0.01218
0.02646 -0.99958
TRAN 36.68436 -68.6383349.21587
#AM
ROTA MATRIX -0.30892 0.95102 -0.01113 0.95109 0.30890 -0.00370
-0.8 -0.01173 -0.3
TRAN 109.48987 -79.0555052.39334
#AN
ROTA MATRIX -0.93676 0.34855 -0.03147 0.34600 0.93589 0.06627 0.05255
0.05119 -0.99731
TRAN 162.32979 -29.2680045.41564
eof
The commonest error with averaging is getting the mask wrong.
Check that the CCs after application of the averaging start at a
reasonable value - 0.3 at least and increase with each cycle ( by the
way why do ncycle 1?)
But in the end the density will not be identical, the Fobs are not
perfectly symmetric so there will be differences. The best idea is to
average (with correct matrices - I always find that takes several pases
before I get them all right - then build molecule A and refit it over
the others before starting refinement.
EleanorHailiang Zhang wrote:
Hi,
I am using the following DM script to perform a NCS averaging. I have a
fundemental question: after NCS averaging, are the density distrubitions
of different NCS unit being averaged supposed to be the same? I found
they
are different by checking FCDM/PHICDM, and maybe I am wrong somewhere...
dm NCSIN ${PDB}.msk HKLIN ${PDBALL}.mtz HKLOUT ${PDBALL}-dm.mtz \
dmtest
mode AVER
ncycle 1
combine PERT
scheme ALL
solc 0.6213
#Identical
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
TRAN 0.0 0.0 0.0
#AC
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.22748 0.97259 0.04813 -0.97372 -0.22662 -0.02273
-0.01120 -0.05203 0.99858
TRAN 101.4683781.74413 2.89341
#AD
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.90337 0.42792 0.02831 -0.42883 -0.90066 -0.07011
-0.00451 -0.07547 0.99714
TRAN 158.0331736.91842 3.25853
#AF
AVER REFI
NCSMASK NMER 1
Re: [ccp4bb] DM NCS averaging question
The commonest error with averaging is getting the mask wrong.
Check that the CCs after application of the averaging start at a
reasonable value - 0.3 at least and increase with each cycle ( by the
way why do ncycle 1?)
But in the end the density will not be identical, the Fobs are not
perfectly symmetric so there will be differences. The best idea is to
average (with correct matrices - I always find that takes several pases
before I get them all right - then build molecule A and refit it over
the others before starting refinement.
EleanorHailiang Zhang wrote:
Hi,
I am using the following DM script to perform a NCS averaging. I have a
fundemental question: after NCS averaging, are the density distrubitions
of different NCS unit being averaged supposed to be the same? I found they
are different by checking FCDM/PHICDM, and maybe I am wrong somewhere...
dm NCSIN ${PDB}.msk HKLIN ${PDBALL}.mtz HKLOUT ${PDBALL}-dm.mtz \
dmtest
mode AVER
ncycle 1
combine PERT
scheme ALL
solc 0.6213
#Identical
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
TRAN 0.0 0.0 0.0
#AC
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.22748 0.97259 0.04813 -0.97372 -0.22662 -0.02273
-0.01120 -0.05203 0.99858
TRAN 101.4683781.74413 2.89341
#AD
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.90337 0.42792 0.02831 -0.42883 -0.90066 -0.07011
-0.00451 -0.07547 0.99714
TRAN 158.0331736.91842 3.25853
#AF
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.21272 -0.97702 0.01352 0.97675 -0.21300 -0.02424 0.02657
0.00805 0.99961
TRAN 100.69797 -81.01860-1.71365
#AG
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.61704 -0.78687 -0.01005 0.78590 0.61553 0.05899 -0.04023
-0.04429 0.99821
TRAN 32.50667 -65.76570 7.05504
#AH
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.85814 -0.51116 -0.04813 -0.51290 0.85771 0.03558
0.02310 0.05522 -0.99821
TRAN 156.1498142.2387348.93406
#AI
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.13703 -0.98975 -0.04032 -0.99048 0.13635 0.01902
-0.01332 0.04254 -0.99901
TRAN 95.9763082.3094852.82510
#AJ
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.69662 -0.71695 -0.02645 -0.71716 -0.69691 0.00230
-0.02008 0.01737 -0.99965
TRAN 25.8458859.7628653.68224
#AK
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.99467 0.10258 -0.01072 0.10259 -0.99472 0.00088 -0.01057
-0.00197 -0.4
TRAN0.47215-8.8608252.78315
#AL
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.55782 0.82946 0.02875 0.82987 -0.55793 -0.00466 0.01218
0.02646 -0.99958
TRAN 36.68436 -68.6383349.21587
#AM
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.30892 0.95102 -0.01113 0.95109 0.30890 -0.00370
-0.8 -0.01173 -0.3
TRAN 109.48987 -79.0555052.39334
#AN
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.93676 0.34855 -0.03147 0.34600 0.93589 0.06627 0.05255
0.05119 -0.99731
TRAN 162.32979 -29.2680045.41564
LABIN FP = FWT PHIO = PHIC FOMO = WCMB
LABOUT FDM=FDM PHIDM=PHIDM FOMDM=FOMDM FCDM=FCDM PHICDM=PHICDM
END
dmtest
Re: [ccp4bb] DM NCS averaging question
Hi,
Thanks for reminding me checking the mask. I think their might be
something wrong with the mask, since when DM read in the mask, it says:
Number of columns, rows, sections ... 84 74 69
Map mode 0
Start and stop points on columns, rows, sections -53 30
80 153 -4 64
Grid sampling on x, y, z 136 260 150
Cell dimensions . 135.57100
260.11200 150.2 90.0 101.14000 90.0
Fast, medium, slow axes .ZXY
Minimum density . 0.0
Maximum density . 0.0
Mean density 0.0
Rms deviation from mean density . 0.0
Space-group .4
Number of titles 1
It seems the mask is just null. However, I converted it to a map file, and
coot clearly showed the mask, so I am not sure why the null mask was found
by DM. Moreover, the NCS CCs are just 0s for the mask.
Anyway, following is my NCSMASK script I used to generate the above mask,
where XYZIN is the reorganized pdb containing only a single fixed NCS unit
(chain A). and all the operations were generated by LSQKAB with Chain A
mapped to other chains. Not sure whether there is something wrong here or
not...
ncsmask xyzin ${PDB}_A.pdb mskout ${PDB}.msk eof
SYMM P1211
EXPAND 1.0
OVERLAP 3
AVERAGE 12
#Identical
ROTA MATRIX 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
TRAN 0.0 0.0 0.0
#AC
ROTA MATRIX -0.22748 0.97259 0.04813 -0.97372 -0.22662 -0.02273
-0.01120 -0.05203 0.99858
TRAN 101.4683781.74413 2.89341
#AD
ROTA MATRIX -0.90337 0.42792 0.02831 -0.42883 -0.90066 -0.07011
-0.00451 -0.07547 0.99714
TRAN 158.0331736.91842 3.25853
#AF
ROTA MATRIX -0.21272 -0.97702 0.01352 0.97675 -0.21300 -0.02424 0.02657
0.00805 0.99961
TRAN 100.69797 -81.01860-1.71365
#AG
ROTA MATRIX 0.61704 -0.78687 -0.01005 0.78590 0.61553 0.05899 -0.04023
-0.04429 0.99821
TRAN 32.50667 -65.76570 7.05504
#AH
ROTA MATRIX -0.85814 -0.51116 -0.04813 -0.51290 0.85771 0.03558
0.02310 0.05522 -0.99821
TRAN 156.1498142.2387348.93406
#AI
ROTA MATRIX -0.13703 -0.98975 -0.04032 -0.99048 0.13635 0.01902
-0.01332 0.04254 -0.99901
TRAN 95.9763082.3094852.82510
#AJ
ROTA MATRIX 0.69662 -0.71695 -0.02645 -0.71716 -0.69691 0.00230
-0.02008 0.01737 -0.99965
TRAN 25.8458859.7628653.68224
#AK
ROTA MATRIX 0.99467 0.10258 -0.01072 0.10259 -0.99472 0.00088 -0.01057
-0.00197 -0.4
TRAN0.47215-8.8608252.78315
#AL
ROTA MATRIX 0.55782 0.82946 0.02875 0.82987 -0.55793 -0.00466 0.01218
0.02646 -0.99958
TRAN 36.68436 -68.6383349.21587
#AM
ROTA MATRIX -0.30892 0.95102 -0.01113 0.95109 0.30890 -0.00370
-0.8 -0.01173 -0.3
TRAN 109.48987 -79.0555052.39334
#AN
ROTA MATRIX -0.93676 0.34855 -0.03147 0.34600 0.93589 0.06627 0.05255
0.05119 -0.99731
TRAN 162.32979 -29.2680045.41564
eof
The commonest error with averaging is getting the mask wrong.
Check that the CCs after application of the averaging start at a
reasonable value - 0.3 at least and increase with each cycle ( by the
way why do ncycle 1?)
But in the end the density will not be identical, the Fobs are not
perfectly symmetric so there will be differences. The best idea is to
average (with correct matrices - I always find that takes several pases
before I get them all right - then build molecule A and refit it over
the others before starting refinement.
EleanorHailiang Zhang wrote:
Hi,
I am using the following DM script to perform a NCS averaging. I have a
fundemental question: after NCS averaging, are the density distrubitions
of different NCS unit being averaged supposed to be the same? I found
they
are different by checking FCDM/PHICDM, and maybe I am wrong somewhere...
dm NCSIN ${PDB}.msk HKLIN ${PDBALL}.mtz HKLOUT ${PDBALL}-dm.mtz \
dmtest
mode AVER
ncycle 1
combine PERT
scheme ALL
solc 0.6213
#Identical
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
TRAN 0.0 0.0 0.0
#AC
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.22748 0.97259 0.04813 -0.97372 -0.22662 -0.02273
-0.01120 -0.05203 0.99858
TRAN 101.4683781.74413 2.89341
#AD
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.90337 0.42792 0.02831 -0.42883 -0.90066 -0.07011
-0.00451 -0.07547 0.99714
TRAN 158.0331736.91842 3.25853
#AF
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.21272 -0.97702 0.01352 0.97675 -0.21300 -0.02424
0.02657
0.00805 0.99961
TRAN 100.69797 -81.01860-1.71365
#AG
AVER
Re: [ccp4bb] DM NCS averaging question
Regarding your script:
change a couple of things:
MODE HIST SOLV MULT AVER
COMBINE PERT
SCHEME RES FROM 3.0 (or from where you have FOM 70%, as your low res phases
will be more reliable)
NCYCLE 50 (since you have 12 molecules you should get a significant benefit
from averaging.
Then the first Aver card does not require the REFI card as it should not be
refined. For all the others I would change them to AVER REFI EVERY 3 to update
the matrices while performing the averaging.
You could also update the solvent mask with the command SOLMASK UPDATE 20 if
you want.
Good luck,
Jürgen
-
Jürgen Bosch
Johns Hopkins Bloomberg School of Public Health
Department of Biochemistry Molecular Biology
Johns Hopkins Malaria Research Institute
615 North Wolfe Street, W8708
Baltimore, MD 21205
Phone: +1-410-614-4742
Lab: +1-410-614-4894
Fax: +1-410-955-3655
http://web.mac.com/bosch_lab/
On Aug 28, 2010, at 10:05 PM, Hailiang Zhang wrote:
Hi,
I am using the following DM script to perform a NCS averaging. I have a
fundemental question: after NCS averaging, are the density distrubitions
of different NCS unit being averaged supposed to be the same? I found they
are different by checking FCDM/PHICDM, and maybe I am wrong somewhere...
dm NCSIN ${PDB}.msk HKLIN ${PDBALL}.mtz HKLOUT ${PDBALL}-dm.mtz \
dmtest
mode AVER
ncycle 1
combine PERT
scheme ALL
solc 0.6213
#Identical
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
TRAN 0.0 0.0 0.0
#AC
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.22748 0.97259 0.04813 -0.97372 -0.22662 -0.02273
-0.01120 -0.05203 0.99858
TRAN 101.4683781.74413 2.89341
#AD
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.90337 0.42792 0.02831 -0.42883 -0.90066 -0.07011
-0.00451 -0.07547 0.99714
TRAN 158.0331736.91842 3.25853
#AF
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.21272 -0.97702 0.01352 0.97675 -0.21300 -0.02424 0.02657
0.00805 0.99961
TRAN 100.69797 -81.01860-1.71365
#AG
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.61704 -0.78687 -0.01005 0.78590 0.61553 0.05899 -0.04023
-0.04429 0.99821
TRAN 32.50667 -65.76570 7.05504
#AH
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.85814 -0.51116 -0.04813 -0.51290 0.85771 0.03558
0.02310 0.05522 -0.99821
TRAN 156.1498142.2387348.93406
#AI
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.13703 -0.98975 -0.04032 -0.99048 0.13635 0.01902
-0.01332 0.04254 -0.99901
TRAN 95.9763082.3094852.82510
#AJ
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.69662 -0.71695 -0.02645 -0.71716 -0.69691 0.00230
-0.02008 0.01737 -0.99965
TRAN 25.8458859.7628653.68224
#AK
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.99467 0.10258 -0.01072 0.10259 -0.99472 0.00088 -0.01057
-0.00197 -0.4
TRAN0.47215-8.8608252.78315
#AL
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.55782 0.82946 0.02875 0.82987 -0.55793 -0.00466 0.01218
0.02646 -0.99958
TRAN 36.68436 -68.6383349.21587
#AM
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.30892 0.95102 -0.01113 0.95109 0.30890 -0.00370
-0.8 -0.01173 -0.3
TRAN 109.48987 -79.0555052.39334
#AN
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.93676 0.34855 -0.03147 0.34600 0.93589 0.06627 0.05255
0.05119 -0.99731
TRAN 162.32979 -29.2680045.41564
LABIN FP = FWT PHIO = PHIC FOMO = WCMB
LABOUT FDM=FDM PHIDM=PHIDM FOMDM=FOMDM FCDM=FCDM PHICDM=PHICDM
END
dmtest
Re: [ccp4bb] DM NCS averaging question
Thanks. I will try the comprehensive script. Originally I just wanted to
see what happens if I do NCS averaging only without solvent flattening or
histogram match, to see whether all NCS units are the same. Anyway, it is
always better to use more of them.
Also thanks for pointing out that the REFI is not necessary for the
identity op.
Best Regards, Hailiang
Regarding your script:
change a couple of things:
MODE HIST SOLV MULT AVER
COMBINE PERT
SCHEME RES FROM 3.0 (or from where you have FOM 70%, as your low res
phases will be more reliable)
NCYCLE 50 (since you have 12 molecules you should get a significant
benefit from averaging.
Then the first Aver card does not require the REFI card as it should not
be refined. For all the others I would change them to AVER REFI EVERY 3 to
update the matrices while performing the averaging.
You could also update the solvent mask with the command SOLMASK UPDATE 20
if you want.
Good luck,
Jürgen
-
Jürgen Bosch
Johns Hopkins Bloomberg School of Public Health
Department of Biochemistry Molecular Biology
Johns Hopkins Malaria Research Institute
615 North Wolfe Street, W8708
Baltimore, MD 21205
Phone: +1-410-614-4742
Lab: +1-410-614-4894
Fax: +1-410-955-3655
http://web.mac.com/bosch_lab/
On Aug 28, 2010, at 10:05 PM, Hailiang Zhang wrote:
Hi,
I am using the following DM script to perform a NCS averaging. I have a
fundemental question: after NCS averaging, are the density distrubitions
of different NCS unit being averaged supposed to be the same? I found
they
are different by checking FCDM/PHICDM, and maybe I am wrong somewhere...
dm NCSIN ${PDB}.msk HKLIN ${PDBALL}.mtz HKLOUT ${PDBALL}-dm.mtz \
dmtest
mode AVER
ncycle 1
combine PERT
scheme ALL
solc 0.6213
#Identical
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
TRAN 0.0 0.0 0.0
#AC
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.22748 0.97259 0.04813 -0.97372 -0.22662 -0.02273
-0.01120 -0.05203 0.99858
TRAN 101.4683781.74413 2.89341
#AD
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.90337 0.42792 0.02831 -0.42883 -0.90066 -0.07011
-0.00451 -0.07547 0.99714
TRAN 158.0331736.91842 3.25853
#AF
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.21272 -0.97702 0.01352 0.97675 -0.21300 -0.02424
0.02657
0.00805 0.99961
TRAN 100.69797 -81.01860-1.71365
#AG
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.61704 -0.78687 -0.01005 0.78590 0.61553 0.05899
-0.04023
-0.04429 0.99821
TRAN 32.50667 -65.76570 7.05504
#AH
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.85814 -0.51116 -0.04813 -0.51290 0.85771 0.03558
0.02310 0.05522 -0.99821
TRAN 156.1498142.2387348.93406
#AI
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.13703 -0.98975 -0.04032 -0.99048 0.13635 0.01902
-0.01332 0.04254 -0.99901
TRAN 95.9763082.3094852.82510
#AJ
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.69662 -0.71695 -0.02645 -0.71716 -0.69691 0.00230
-0.02008 0.01737 -0.99965
TRAN 25.8458859.7628653.68224
#AK
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.99467 0.10258 -0.01072 0.10259 -0.99472 0.00088
-0.01057
-0.00197 -0.4
TRAN0.47215-8.8608252.78315
#AL
AVER REFI
NCSMASK NMER 1
ROTA MATRIX 0.55782 0.82946 0.02875 0.82987 -0.55793 -0.00466 0.01218
0.02646 -0.99958
TRAN 36.68436 -68.6383349.21587
#AM
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.30892 0.95102 -0.01113 0.95109 0.30890 -0.00370
-0.8 -0.01173 -0.3
TRAN 109.48987 -79.0555052.39334
#AN
AVER REFI
NCSMASK NMER 1
ROTA MATRIX -0.93676 0.34855 -0.03147 0.34600 0.93589 0.06627
0.05255
0.05119 -0.99731
TRAN 162.32979 -29.2680045.41564
LABIN FP = FWT PHIO = PHIC FOMO = WCMB
LABOUT FDM=FDM PHIDM=PHIDM FOMDM=FOMDM FCDM=FCDM PHICDM=PHICDM
END
dmtest
