Previous Thread
Next Thread
Print Thread
problem of PBC with crystal: need help
#10197 04/09/06 06:25 PM
Joined: Jan 2006
Posts: 22
T
tong Offline OP
Forum Member
OP Offline
Forum Member
T
Joined: Jan 2006
Posts: 22
Dear Charmm users,

I performed periodic boundary condition using crystal command and after equilibration for10 ps, I ]found that my protein was located outside water box(please see the attached file). How should I configure it out? Here is my periodic setup.

!------------------------------------------------------------------------
!Define first PBC:

set ax 67.054
set by 68.118
set cz 112.073


CRYSTAL DEFINE ORTHorhombic @ax @by @cz 90.0 90.0 90.0

! Cutoff may vary depending on your needs

CRYSTAL BUILD CUTOFF @ct nooperation 0

update inbfrq -1 ihbfrq 0 imgfrq 20

! Then defines images:
! Note :
! 1. IMAGE BYSEG is not correct for water or ion segments;
! it only correct if the segment is a whole molucule
! so use image byseg for protein
! 2. Use "image byres" for water, ions, or any case where each
! residue is a complete for a molecular species

IMAGE BYRESID XCEN 0.0 YCEN 0.0 ZCEN 0.0 sele resn TIP3 .or. resn MG .or. resn CIT end

IMAGE BYSEG sele segid PROT end

Any suggestion/comment is appreciated. Thanks in advance.

Attached Files
Re: problem of PBC with crystal: need help
tong #10198 04/09/06 07:10 PM
Joined: Sep 2003
Posts: 8,501
rmv Online Content
Forum Member
Online Content
Forum Member
Joined: Sep 2003
Posts: 8,501
The entire molecule must be completely outside of the box for image centering to occur; only a small part of your protein is outside the box. This is normal, and shouldn't be a problem, assuming the CUTOFF used with CRYSTAL BUILD was large enough (hint: you should have at least 26 image transformations; check your output).

Note that this setup (ORTH lattice) is not advised for constant P simulation with the CPT code; because the pressure tensor is scaled anisotropically, each length can change independently, which can lead to a collapse along one of the dimensions. An isotropic shape (CUBIC, OCTA, RHDO) is highly recommended, esp. the latter two.


Rick Venable
computational chemist

Re: problem of PBC with crystal: need help
rmv #10199 04/09/06 07:56 PM
Joined: Jan 2006
Posts: 22
T
tong Offline OP
Forum Member
OP Offline
Forum Member
T
Joined: Jan 2006
Posts: 22
Dear Prof. Venable,

Thanks so much for your quickly reply. However I still have questions.



1. only a small part of your protein is outside the box. This is normal, and shouldn't be a problem – this issue is still not clear to me. May be I miss understand some points so I appreciate if you can explain me more. I just confuse because I think normally protein should be surrounded by water/solvent molecules. In this case, some part of my protein is outside the box, how it can interact with water molecules (because there is no more water in this area, I mean outside the box)
2. assuming the CUTOFF used with CRYSTAL BUILD was large enough Here is my cutoff setup
set ct 13.0 ! Distance cutoff for generating list of pairs
set ctof 12.0 ! Distance cut at which the switching function eliminates all
! contributions from a pair in calculating energies
set cton 8.0 ! Distance cut at which the smoothing function begins to reduce
! a pair's contribution.

CRYSTAL BUILD CUTOFF @ct nooperation 0

Do you think it is OK?

3. (hint: you should have at least 26 image transformations; check your output). What did you refer to “imgae transformation”? If it refers to this part of my output file (see below), I have more than 26 transformations.
SELECTED IMAGES ATOMS BEING CENTERED ABOUT 0.000000 0.000000 0.000000
RESIDUE 159 OPERATED ON BY TRANSFORMATION C008
RESIDUE 160 OPERATED ON BY TRANSFORMATION C004
RESIDUE 162 OPERATED ON BY TRANSFORMATION C007
RESIDUE 163 OPERATED ON BY TRANSFORMATION C007
RESIDUE 164 OPERATED ON BY TRANSFORMATION C007
RESIDUE 165 OPERATED ON BY TRANSFORMATION C004
RESIDUE 170 OPERATED ON BY TRANSFORMATION C008
RESIDUE 171 OPERATED ON BY TRANSFORMATION C007
RESIDUE 173 OPERATED ON BY TRANSFORMATION C007


Please advice. Thanks in advance.

Re: problem of PBC with crystal: need help
tong #10200 04/09/06 09:22 PM
Joined: Sep 2003
Posts: 8,501
rmv Online Content
Forum Member
Online Content
Forum Member
Joined: Sep 2003
Posts: 8,501
What you have posted are the listing of water residue image centering operations; there are many listings because there are many water molecules near the edge of the box. These are not the image transformations, but the result of their use; the identity of the tranformation is indicated in the the listing by 'C007' etc., however. The actual number of transformations created should be listed immediately after CRYSTAL BUILD.

If done properly, the "real" coordinates (which are the only ones in your images) are surrounded by a minimum of 26 exact "image" copies. These image atoms are used only for computing nonbond forces; their coordinates are determined by real atoms, from an opposing parallel face of the geometric shape for simple lattices with only translational symmetry. As long as you've properly defined the transformations, the protein can be mostly out of the box, but the energy and forces will be correct because it is surrounded by image water molecules. If you examine your box carefully, you should see a concavity in the water on side opposite that where the protein protudes. How to include the image atoms in output coordinate files has been described previously in these forums.

The CUTOFF for CRYSTAL BUILD may be a bit small; it can (and probably should) be quite a bit larger than CUTNB and CUTIM. A value of more like 30 would be more reasonable for a box of this size.

You haven't indicated the value of CUTIM you are using; I usually set CUTIM and CUTNB to the same values, and do the same for IMGFRQ and INBFRQ.


Rick Venable
computational chemist

Re: problem of PBC with crystal: need help
rmv #10201 04/09/06 10:20 PM
Joined: Jan 2006
Posts: 22
T
tong Offline OP
Forum Member
OP Offline
Forum Member
T
Joined: Jan 2006
Posts: 22

Dear Prof. Venable,

Thanks so much for your nice explanation. It’s really helpful. Now I got what the image transformation refers to. I have only 8 image transformations in this case.

Parameter: AX -> "67.054"
Parameter: BY -> "68.118"
Parameter: CZ -> "112.073"
Crystal Parameters : Crystal Type = ORTH
A = 67.05400 B = 68.11800 C = 112.07300
Alpha = 90.00000 Beta = 90.00000 Gamma = 90.00000

CHARMM>

CHARMM> ! Cutoff may vary depending on your needs
CHARMM>

CHARMM> CRYSTAL BUILD CUTOFF @ct nooperation 0
Parameter: CT -> "13.0"
**** Warning **** The following extraneous characters
were found while command processing in XBUILD
NOOPERATION 0

Range of Grid Search for Transformation 1 :
Lattice Vector A -2 TO 2
Lattice Vector B -2 TO 2
Lattice Vector C -1 TO 1


The number of transformations generated = 8


Number Symop A B C Distance

1 1 -1 -1 0 5.5634
2 1 -1 0 0 2.0715
3 1 -1 1 0 5.9206
4 1 0 -1 0 1.7623
5 1 0 1 0 1.7623
6 1 1 1 0 5.5634
7 1 1 0 0 2.0715
8 1 1 -1 0 5.9206
THERE ARE NO ROTATIONS FOR THIS TRANSFORMATION SET
8 Transformations have been processed.


Yes, in this case I did not define CUTIM so I think I will define it and try to include the image atoms in out put coordinate files. One more question is that is it necessary to use the same values for “cutnb: and cutoff for crystal build. Thanks in advance.

Re: problem of PBC with crystal: need help
tong #10202 04/09/06 10:25 PM
Joined: Jan 2006
Posts: 22
T
tong Offline OP
Forum Member
OP Offline
Forum Member
T
Joined: Jan 2006
Posts: 22

Dear Prof. Venable,

Sorry for the last post. The answer was already indicated in your previous massage. Thanks.

Re: problem of PBC with crystal: need help
tong #10203 04/10/06 06:48 PM
Joined: Jan 2006
Posts: 22
T
tong Offline OP
Forum Member
OP Offline
Forum Member
T
Joined: Jan 2006
Posts: 22
Dear Prof. Venable,

I still have problem with PBC set up.

1. After I changed the ORTH lattice to OCTA and RHDO, as you suggested, I found that both lattice types are not suitable for my system. I’ve found massage “XUCELL Symmetry is not conserved”, in my output file. I’m not sure what’s wrong in my input file (I think it should be fine).


! First define PBC:

set ax 67.054
set by 68.118
set cz 112.073

! Here select OCTA
CRYSTAL DEFINE OCTA @ax @by @cz 109.5 109.5 109.5

CRYSTAL BUILD CUTOFF 30 nooperation 0

! use the sane for inbfrq and imgfrq
update inbfrq 50 ihbfrq 0 imgfrq 50 ! -1 : update when necessary
! 50 = default

IMAGE BYRESID XCEN 0.0 YCEN 0.0 ZCEN 0.0 sele resn TIP3 .or. resn MG .or. resn CIT end

IMAGE BYSEG sele segid PROT end


shake bonh

! non-bond specifications
! see nbonds.doc for details
NBOND cutnb @ct ctofnb @ctof ctonnb @cton cutim @ctim -
atom fswitch cdie vdw vshift wmin 1.5 esp 1.0

Output:


CHARMM> ! CUBIC- a=b=c and alpha = beta = gamma = 90 degrees
CHARMM> ! OCTA- a=b=c & alpha = beta = gamma = 109.5 degrees
CHARMM> ! RHDO-a=b=c, alpha = gamma =60 degrees and beta = 90.0 degrees
CHARMM>

CHARMM>

CHARMM> ! Here select OCTA

CHARMM> CRYSTAL DEFINE OCTA @ax @by @cz 109.5 109.5 109.5
Parameter: AX -> "67.054"
Parameter: AX -> "67.054"
***** XUCELL Symmetry is not conserved
******************************************
BOMLEV ( -3) IS REACHED - TERMINATING. WRNLEV IS -5

ABNORMAL TERMINATION
MAXIMUM STACK SPACE USED IS 163196
STACK CURRENTLY IN USE IS 0
MOST SEVERE WARNING WAS AT LEVEL -4

$$$$$ JOB ACCOUNTING INFORMATION $$$$$
ELAPSED TIME: 2.12 SECONDS
CPU TIME: 0.00 SECONDS
p3_29996: p4_error: : 12
***** LEVEL -4 WARNING FROM <XTLSYM> *****
***** XUCELL Symmetry is not conserved
******************************************
BOMLEV ( -3) IS REACHED - TERMINATING. WRNLEV IS -5


Output of Rhdo:


CHARMM> CRYSTAL DEFINE RHDO @ax @by @cz 60.0 90.0 60.0
Parameter: AX -> "67.054"
Parameter: BY -> "68.118"
Parameter: CZ -> "112.073"

***** LEVEL -4 WARNING FROM <XTLSYM> *****
***** XUCELL Symmetry is not conserved


As you mentioned that ORTH is not advised for CPT MD so I’m not sure whether it’s reasonable if I still use ORTH for CPT MD.

2. The another issue is that I still have only 8 image transformations, as shown below.
CHARMM> CRYSTAL BUILD CUTOFF 30 nooperation 0
**** Warning **** The following extraneous characters
were found while command processing in XBUILD
NOOPERATION 0

Range of Grid Search for Transformation 1 :
Lattice Vector A -2 TO 2
Lattice Vector B -2 TO 2
Lattice Vector C -1 TO 1


The number of transformations generated = 8


Number Symop A B C Distance

1 1 -1 -1 0 5.5634
2 1 -1 0 0 2.0715
3 1 -1 1 0 5.9206
4 1 0 -1 0 1.7623
5 1 0 1 0 1.7623
6 1 1 1 0 5.5634
7 1 1 0 0 2.0715
8 1 1 -1 0 5.9206
THERE ARE NO ROTATIONS FOR THIS TRANSFORMATION SET
8 Transformations have been processed.

How can I increase this value up to 26 image transformations, as your above massage?


3. If I’d like to bring my protein back to the box, how should I do? I’ve searched about this issue and I found one massage, like this:

3) It is found that there are water molecules are outside of the box even if crystal/cpt used. is there anything wrong with the running? If this is just Ok, then how to bring everything back into the box.

CRYSTAL and CPT don't handle that, you must set up the IMAGE centering commands as well, based on your system topology. If water molecules, etc., drift too far from the central box it is a problem.

Is “set up the IMAGE centering” refers to these lines in my input?

IMAGE BYRESID XCEN 0.0 YCEN 0.0 ZCEN 0.0 sele resn TIP3 .or. resn MG .or. resn CIT end

IMAGE BYSEG XCEN 0.0 YCEN 0.0 ZCEN 0.0 sele segid PROT end

If that, why my protein is still outside the box(see attached file)? If not, please provide me more detail how to bring protein back to water box. Thanks in advance.

Attached Files
9941-AfterequilibrationusingPBC.doc (0 Bytes, 459 downloads)
Re: problem of PBC with crystal: need help
tong #10204 04/10/06 08:16 PM
Joined: Sep 2003
Posts: 8,501
rmv Online Content
Forum Member
Online Content
Forum Member
Joined: Sep 2003
Posts: 8,501
For OCTA and RHDO, all the lengths must be equal; hence the message "XUCELL Symmetry is not conserved", as you've specified unequal lengths. OCTA and RHDO are sphere-like shapes, not boxes with square corners, so you'd likely need to resolvate your system in a more spherical shape.

IMAGE BYRES and IMAGE BYSEG are the image centering commands. As I've already noted above, the protein position will not be subject to image centering during a simulation until all of its atoms are completely outside the unit cell. There are processing options (see MERGE in dynamc.doc) to RECENTER the solute after the simulation is finished.

While at least 26 transformations are needed for box-like shapes, the minimum number needed is different for lattice shapes with more than 6 faces, or for more complex symmetry with rotations. A general strategy is to increase the CUTOFF value used for CRYSTAL BUILD until a subsequent ENERGY evaluation doesn't change. If you increase the CUTOFF and the total ENERGY changes, you didn't have enough transformations defined.

As for using ORTH with DYNA CPT for NPT dynamics, I've heard several reports of collapse along one dimension, and had it happen to me once as well. I strongly advise against the ORTH lattice for constant P simulations.


Rick Venable
computational chemist

Re: problem of PBC with crystal: need help
rmv #10205 04/17/06 04:59 PM
Joined: Jan 2006
Posts: 22
T
tong Offline OP
Forum Member
OP Offline
Forum Member
T
Joined: Jan 2006
Posts: 22
Dear Prof. Venable,

Thanks so much for your explanation. Now my simulation using PBC works well.


Moderated by  BRBrooks, lennart, rmv 

Link Copied to Clipboard
Powered by UBB.threads™ PHP Forum Software 7.7.4
(Release build 20200307)
Responsive Width:

PHP: 5.6.33-0+deb8u1 Page Time: 0.009s Queries: 34 (0.005s) Memory: 0.9717 MB (Peak: 1.0974 MB) Data Comp: Off Server Time: 2020-10-21 05:15:24 UTC
Valid HTML 5 and Valid CSS