Topic Options
#14901 - 07/04/07 05:45 PM JOIN command
Chaitanya Offline
Forum Member

Registered: 06/12/07
Posts: 21
Loc: Kanpur, India
Hello,

I have a basic doubt. What is the exact function of the JOIN command? Does it 'join' i.e create any linkage(bond) between two segments? Can the command be used for two subunits of a protein which are not covalently connected to each other?I read the struc.doc, but am not clear about its use.

Thanks in advance..

Top
#14902 - 07/04/07 07:03 PM Re: JOIN command [Re: Chaitanya]
rmv Online   content

Forum Member

Registered: 09/17/03
Posts: 8373
Loc: 39 03 48 N, 77 06 54 W
The JOIN command has two main uses--
  • It combine two segments into one segment; the segments must be adjoining, i.e. consecutive in the PSF. No new bonds are created, the PATCH command is typically used for that.
  • It can RENUMber the RESIDs, so that they are consecutive; useful when combining as above, but can also applied to a water segment after deletion of overlapping waters.

Top
#14903 - 07/05/07 06:40 AM Re: JOIN command [Re: rmv]
Chaitanya Offline
Forum Member

Registered: 06/12/07
Posts: 21
Loc: Kanpur, India
Thanks Rick

Top
#14904 - 09/04/07 05:22 PM Re: JOIN command [Re: rmv]
preeti Offline
Forum Member

Registered: 01/05/05
Posts: 127
Hello,

I have two different chains of a protein that are not colavently linked. I read the coordinates for two chains from files using OFFSET command.

Then, in order to renumber the two chain residues continuously by trying to keep chain A beginning from 1 to 90 and chain B beginning from 91 to 270, i used:

join A B renum

The result is the final pdb which is renumbered but when I view the renumbered PDb in VMD, I see the end residues (resid 90 and resid 91) are displaced from their original position and it seems the JOIN command is leading them to form a covalent bond where the residues begin to renumber [90 (chainA) and 91 (chain B)].

Am I doing something wrong?
To get rid of above problem, I used:

join A renum
! which renums only chain A

Is there any way to renumber chain B starting from 91 using any command other than join?

Thanks, preeti

Top
#14905 - 09/04/07 06:17 PM Re: JOIN command [Re: preeti]
rmv Online   content

Forum Member

Registered: 09/17/03
Posts: 8373
Loc: 39 03 48 N, 77 06 54 W
You should not necessarily regard a line drawn on the screen by VMD as proof that CHARMM has defined a bond between the two segments in this case.

If you used separate GENErate commands for each segment with the appropriate end patches, there should not be a bond connecting the segments in the CHARMM PSF after using the JOIN command to combine the two segments.

Top
#14906 - 09/05/07 01:29 PM Re: JOIN command [Re: rmv]
preeti Offline
Forum Member

Registered: 01/05/05
Posts: 127
Rick,

I realized that join command is not leading to any error.

I see that the coordinates of last residue (resid 90-chain A) and (resid 91-chain B) end up shifting towards each other on minimizatin in vacuum even with a constraint on the backbone atoms. I do not why this is happening at all. Further when i minimize using GBMV to release the constraint slowly from 5 to 1 "units", the residues end up forming a chain between two chians, which is obviously wrong.

The initial structure upon reading and joining two chains (nat-ab.pdb) looks fine when i compare to the original pdb.

Could you help me identify the reasons from my script, why could this happening? Is it becuase I am trying to use "ace" and "ct3" as caps for end terminals in both chains which in some way is leading for the two ends to be attracted to each other? or because of nonbond parameters?

Thanks much,
Preeti

! script for reading in two chains:

open unit 1 read card name nata.pdb
read sequ pdb unit 1
generate A setup warn first ace last ct3

read COORDINATES pdb UNIT 1 offset -15
CLOSE UNIT 1

ic purge
ic param
ic fill preserve
ic build
define test sele (.not. type H*) .and. -
(.not. init) show end

coor init sele type H* end
hbuild sele type H* end
define test sele .not. init show end

open unit 1 read card name natb.pdb
read sequ pdb unit 1
rewind unit 1
generate B setup warn first ace last ct3

read COORDINATES pdb UNIT 1 offset -340
CLOSE UNIT 1

ic purge
ic param
ic fill preserve
ic build
define test sele (.not. type H*) .and. -
(.not. init) show end

coor init sele type H* end
hbuild sele type H* end
define test sele .not. init show end

join A B renum

open unit 1 write card name nat-ab.pdb
write COORDINATES pdb UNIT 1
CLOSE UNIT 1

stop

Then I run the minimization in vacuum:
--------------------------------------

open unit 1 read card name nat-ab.pdb
read sequ pdb unit 1
close unit 1
generate AB setup warn first ace last ct3

open unit 1 read card name nat-ab.crd
read COORDINATES card UNIT 1
CLOSE UNIT 1

set ctofnb = 16.0
set ctonnb = @ctofnb
calc cutnb = @ctofnb + 4.0

cons harm force 10.0 mass select -
( type c .or. type o .or. type ca .or. type n ) end

mini sd nstep 100 tolgrd 0.0001 update ctonnb @ctonnb ctofnb @ctofnb atom vatom vswi vdistance cdie cutnb @cutnb EPS 1.0 E14FAC 1.0 wmin 1.0

mini abnr nstep 200000 tolgrd 0.0001 update ctonnb @ctonnb - ctofnb @ctofnb atom vatom vswi vdistance cdie cutnb @cutnb EPS 1.0 E14FAC 1.0 wmin 1.0

open write unit 10 card name nat-ab-mini.pdb
write coor unit 10 pdb
* mini ab
*
close unit 10

stop

Top
#14907 - 09/06/07 05:20 PM Re: JOIN command [Re: preeti]
rmv Online   content

Forum Member

Registered: 09/17/03
Posts: 8373
Loc: 39 03 48 N, 77 06 54 W
Try a shorter minimization; 200000 steps is actually quite a lot.

The CTONNB value is problematic; it should be 3-4 A shorter than CTOFNB.

Typical values are more like CUTNB 16. CTOFNB 12. CTONNB 8.; most recent parameter sets were developed with these CTOFNB and CTONNB values, and changing them substantially amounts to changing the force field.

If a shorter minimization with more appropriate CTOFNB and CTONNB values does not help much, try increasing the restraint force constant; if the two fragments are fairly close (<10 A) the non-bond interactions would tend to pull them together (esp. with longer cutoffs).

Top
#14908 - 09/07/07 10:37 AM Re: JOIN command [Re: rmv]
preeti Offline
Forum Member

Registered: 01/05/05
Posts: 127
Hello Rick,

Thank you for the suggestions. I tried them all your suggestions, none of them affects the attraction between the two end points. I tried minimization separately both in vacuum and GBMV implicit model. Could the problem with that one chain of my protein is a very long helix which is not stable in the vacuum/implicit solvent. The distance between the two ends (resid 90 chain A and resid 91 chain B) is 50 angstroms.

Is the only solution is putting it in water box to prevent the helix to not loose its helix character.

preeti

Top
#14909 - 09/07/07 11:29 AM Re: JOIN command [Re: preeti]
rmv Online   content

Forum Member

Registered: 09/17/03
Posts: 8373
Loc: 39 03 48 N, 77 06 54 W
Make 3 copies of the minimization script, put one of the following lines in each, and run the 3 scripts:

SKIPE VDW
SKIPE ELEC
SKIPE BOND

Change the harmonic restraints to include all heavy atoms, e.g.

CONS HARM ABSO FORCE 10. MASS SELE NOT HYDROGEN END

Omit the H atoms from graphics display (esp. for SKIPE BOND). Hopefully, this should indicate which energy term is contributing the most to the attractive forces between the molecules.

Top

Moderator:  John Legato, lennart