Patches and cofactors in charmm

[acellera]

Patches and cofactors in charmm

[noeliaferruz]

hahah yes! Exactly, resname PTR and not backbone removes the sidechain, and one oxygen of the phosphate. It keeps the backbone and the phosphate, and two of its three oxygen. xD. amazing.

[mj-harvey]

Adreasing your last point, I'm not confident that parametrise in its
current state can correctly parameterize a modified residue. The method's
in principle capable of it, though, as evidenced by the gammp paper. Let me
take a look over the weekend and see.
M
On 24 Mar 2016 15:47, "Noelia Ferruz" [email protected] wrote:

Hello guys,

I am not finding an easy way to add patches or parameterize cofactors
whose parameters are included in charmm already (like phosphates (TP2),
hemoglobine (HEME), INO1.. etc)

The problem comes from the fact that these RESIs are (only) included in
the charmm_27prot_na ff. This has two consequences:

1. (probably) many of the most standard patches and cofactors haven't
   been reparameterized since the release of C27.
2. Adding patches using the lastest ff requieres sourcing the ff of
   the patch (RESI), which also depends on the charmm27na ff atomtypes. You
   can of course separate the RESI you want to apply along with the atomtypes
   it depends on (select those manually, and are different depending on the
   residue) in a separate file.

I was wondering if this workaround in 2), which is error-prone and
requires a lot of manual intervention could be easily handled with htmd.
Phosphates, hemoglobines and other molecules are present in everyday
simulations.
The other question is if these RESIs wouldnt be better described with
parameterise.

Let me know what you think.

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub
#3

[giadefa]

We have done it already
On 24 Mar 2016 22:35, "M J Harvey" [email protected] wrote:

Adreasing your last point, I'm not confident that parametrise in its
current state can correctly parameterize a modified residue. The method's
in principle capable of it, though, as evidenced by the gammp paper. Let me
take a look over the weekend and see.
M
On 24 Mar 2016 15:47, "Noelia Ferruz" [email protected] wrote:

Hello guys,

I am not finding an easy way to add patches or parameterize cofactors
whose parameters are included in charmm already (like phosphates (TP2),
hemoglobine (HEME), INO1.. etc)

The problem comes from the fact that these RESIs are (only) included in
the charmm_27prot_na ff. This has two consequences:

1. (probably) many of the most standard patches and cofactors haven't
   been reparameterized since the release of C27.
2. Adding patches using the lastest ff requieres sourcing the ff of
   the patch (RESI), which also depends on the charmm27na ff atomtypes. You
   can of course separate the RESI you want to apply along with the
   atomtypes
   it depends on (select those manually, and are different depending on the
   residue) in a separate file.

I was wondering if this workaround in 2), which is error-prone and
requires a lot of manual intervention could be easily handled with htmd.
Phosphates, hemoglobines and other molecules are present in everyday
simulations.
The other question is if these RESIs wouldnt be better described with
parameterise.

Let me know what you think.

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub
#3

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub
#3 (comment)

[mj-harvey]

Which 'it'?
On 25 Mar 2016 09:23, "giadefa" [email protected] wrote:

We have done it already
On 24 Mar 2016 22:35, "M J Harvey" [email protected] wrote:

Adreasing your last point, I'm not confident that parametrise in its
current state can correctly parameterize a modified residue. The method's
in principle capable of it, though, as evidenced by the gammp paper. Let
me
take a look over the weekend and see.
M
On 24 Mar 2016 15:47, "Noelia Ferruz" [email protected] wrote:

Hello guys,

I am not finding an easy way to add patches or parameterize cofactors
whose parameters are included in charmm already (like phosphates (TP2),
hemoglobine (HEME), INO1.. etc)

The problem comes from the fact that these RESIs are (only) included in
the charmm_27prot_na ff. This has two consequences:

1. (probably) many of the most standard patches and cofactors haven't
   been reparameterized since the release of C27.
2. Adding patches using the lastest ff requieres sourcing the ff of
   the patch (RESI), which also depends on the charmm27na ff atomtypes.
   You
   can of course separate the RESI you want to apply along with the
   atomtypes
   it depends on (select those manually, and are different depending on
   the
   residue) in a separate file.

I was wondering if this workaround in 2), which is error-prone and
requires a lot of manual intervention could be easily handled with
htmd.
Phosphates, hemoglobines and other molecules are present in everyday
simulations.
The other question is if these RESIs wouldnt be better described with
parameterise.

Let me know what you think.

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub
#3

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub
#3 (comment)

—
You are receiving this because you commented.
Reply to this email directly or view it on GitHub
#3 (comment)

[stefdoerr]

I added CHARMM stream files in the new multiscalelab/htmd. Do charmm.listFiles() and check them out since they have heme parametrizations and other interesting stuff.

[noeliaferruz]

Fantastic!! I just saw it.
How does htmd read str files? (As they contain both prm and rtf in one single file, do we have to split them before building?)

[stefdoerr]

Yes, it splits them into some temp files and then copies them into your
build directory. I have tested only the splitting though so no guarantees
the rest of the pipeline works. Do tell me though ;) Should be clear enough
since the files should appear in the build dir.

On Tue, Apr 5, 2016 at 6:17 PM, Noelia Ferruz [email protected]
wrote:

Fantastic!! I just saw it.
How does htmd read str files? (As they contain both prm and rtf in one
single file, do we have to split them before building?)

—
You are receiving this because you commented.
Reply to this email directly or view it on GitHub
#3 (comment)

[giadefa]

pratically they are usable directly by charmmBuilt

On 5 April 2016 at 19:18, Stefan [email protected] wrote:

Yes, it splits them into some temp files and then copies them into your
build directory. I have tested only the splitting though so no guarantees
the rest of the pipeline works. Do tell me though ;) Should be clear enough
since the files should appear in the build dir.

On Tue, Apr 5, 2016 at 6:17 PM, Noelia Ferruz [email protected]
wrote:

Fantastic!! I just saw it.
How does htmd read str files? (As they contain both prm and rtf in one
single file, do we have to split them before building?)

—
You are receiving this because you commented.
Reply to this email directly or view it on GitHub
#3 (comment)

—
You are receiving this because you commented.
Reply to this email directly or view it on GitHub
#3 (comment)

http://www.acellera.com

   <https://twitter.com/acellera>

https://www.youtube.com/user/acelleracom
https://www.linkedin.com/company/2133167?trk=tyah&trkInfo=clickedVertical%3Acompany%2CclickedEntityId%3A2133167%2Cidx%3A2-1-2%2CtarId%3A1448018583204%2Ctas%3Aacellera
https://www.acellera.com/md-simulation-blog-news/
http://is.gd/1eXkbS

[giadefa]

Can this be closed @noeliaferruz ?

[noeliaferruz]

Hola!

Sorry, I hadn't quite tested it. I have just tried to build the case with the phosphate patch. It's nothing important, I'm just trying to see if I can actually build a system with phosphates. These are the lines I'm using:

./1.build.py 5hx8.pdb 8 0 [Where 1.build.py looks like:]

system = Molecule(sys.argv[1])
system.filter("chain A and (protein or water)")
systemCaps = autoSegment(system, "protein", "P")
systemCaps.mutateResidue('resname PTR', 'TYR')
s = float(sys.argv[2])
d = maxDistance( systemCaps, "all" )
d=d+s
solvated=solvate(systemCaps, minmax=[[-d, -d, -d], [d,d,d]])
topos= [ "top/top_all36_prot.rtf",  "top/top_water_ions.rtf"]
params=[ "par/par_all36_prot_mod.prm", "par/par_water_ions.prm", "str/na/toppar_all36_na_model.str"]
built= charmm.build( solvated, topo=topos, param=params, saltconc=float(sys.argv[3]), outdir='build', patches='patch TP2 A:1035')

---------------------------------------------------------------------------------------------------------------
Videos from the HTMD2015 workshops are available on the Acellera youtube channel: https://www.youtube.com/user/acelleralive

You are on the latest HTMD version (1.0.16).
Solvating: 100% (27/27) [#######################################################################################################################################################################################################] eta 00:01 |
Traceback (most recent call last):
  File "./build.py", line 25, in <module>
    built= charmm.build( solvated, topo=topos, param=params, saltconc=float(sys.argv[3]), outdir='build', patches='patch TP2 A:1035')
  File "/nfs/grid/software/hpcc/apps/Linux-x86_64-RHEL6/acellera/current/python/lib/python3.5/site-packages/htmd/builder/charmm.py", line 127, in build
    caps = _defaultCaps(mol)
  File "/nfs/grid/software/hpcc/apps/Linux-x86_64-RHEL6/acellera/current/python/lib/python3.5/site-packages/htmd/builder/charmm.py", line 412, in _defaultCaps
    raise NameError('Segments {} contain both protein and non-protein atoms. Please assign separate segments to them.'.format(intersection))
NameError: Segments ['P2'] contain both protein and non-protein atoms. Please assign separate segments to them.

Visual inspection reveals that the phosphate is still there. I know @stefdoerr fixed this exact issue in Molecule.py some weeks ago, so I don't know.

So, no don't close it yet, I'd like to find a way to add phosphates and other cofactors!

Thanks,
Noelia

[stefdoerr]

No, I think the issue was that VMD atomselect was recognizing phosphates as "protein". Seems to be the case here at least. I don't know a fix for that, other than making your autoSegment atomselection fancier to exclude the phosphates.

[noeliaferruz]

But its mutateResidue who's not removing the entire sidechain of PTR, no?

[stefdoerr]

Ah if that's the case (which should be fixed), just swap the order. First do stuff like mutating etc and then autoSegment. Even if the phosphates are not removed (which they should), they will not be bonded any more to the protein so then the autoSegment should work correct.

[stefdoerr]

You are sure the phosphates belong to the old PTR residues?

[noeliaferruz]

Ok, I'll do that.
Yes, mutateResidue keeps P, OP1 and OP2 from the original PTR residue but renames them to TYR.

[stefdoerr]

Wow. That means that the VMD atomselection recognizes those atoms as backbone. Can you verify that? Because I remove all atoms with 'resname PTR and not backbone'. I somehow assumed this would cover all weird atoms attached to a residue.

[j3mdamas]

Maybe this could be sent to the VMD mailing list?

On Thu, Apr 21, 2016 at 6:08 PM, Noelia Ferruz [email protected]
wrote:

hahah yes! Exactly, resname PTR and not backbone removes the sidechain,
and one oxygen of the phosphate. It keeps the backbone and the phosphate,
and two of its three oxygen. xD. amazing.
[image: screenshot]
https://cloud.githubusercontent.com/assets/12035024/14716171/c4ce9e34-07b9-11e6-9523-134c263ab0e4.png

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub
#3 (comment)

[stefdoerr]

Kill me... :P Well that is somehow beyond my capabilities, haha. If anyone can find a generalized atomselect (that works on all residues) and removes them I will implement it.

[stefdoerr]

Yeah we could try our luck there. @noeliaferruz can you send the residue (and maybe one or two before and after for "context") by mail so that I can ask them?

[noeliaferruz]

The backbone atoms are always N CA C O, no?

[j3mdamas]

I follow their mailing list, it's quite active.

On Thu, Apr 21, 2016 at 6:13 PM, Stefan [email protected] wrote:

Yeah we could try our luck there. @noeliaferruz
https://github.com/noeliaferruz can you send the residue (and maybe one
or two before and after for "context") by mail so that I can ask them?

—
You are receiving this because you commented.
Reply to this email directly or view it on GitHub
#3 (comment)

[stefdoerr]

Hm Noelia, the protein structure was it by any chance written by HTMD at some point? We had a small very weird issue with the PDB writer. Can you send me like the original original pdb structure of this residue?

[stefdoerr]

Apparently in the PDB format there is a significance on where you put atom names. So carbon alphas should be written in columns 13-16 as "..CA" but calcium should be ".CA." where dots indicate spaces. So if VMD assigns significance to that it could have problems.

[noeliaferruz]

The original pdb: 5hx8.pdb
5hx8.txt
The one just wrote with htmd:
protein.txt

** But I'm using the first for the selections in the picture **

[stefdoerr]

Ok thanks. It's not our error. Happens also if you get the PDB from the PDB website directly. Will ask the VMD mailing list.

[noeliaferruz]

Would in the meanwhileresname PTR and not name N CA C O work?

[stefdoerr]

yes, do it manually. Do mol.remove('resname PTR and not name N CA C O') and then mol.set('resname', 'TYR', sel='resname PTR')

[noeliaferruz]

yep.
Continuing...

[stefdoerr]

Ok apparently backbone in VMD works for both protein and nucleic acids. In this case it thought that your P, PO etc were nucleic acid backbone. Seems like I can substitute the backbone selection with C CA N O without loss of generality so I will do that.

[j3mdamas]

Maybe create a new selection protein-backbone instead of substitution?

[stefdoerr]

There is no such atomselection in VMD. I don't think I can invent atomselection shortcuts either.

[stefdoerr]

Not without modifying the VMD source code at least which I don't see as a real need. It would end up doing the same thing anyways.

[j3mdamas]

forget about it. you were talking about your specific application not HTMD
:)

On Fri, Apr 22, 2016 at 11:43 AM, Stefan [email protected] wrote:

Not without modifying the VMD source code at least which I don't see as a
real need. It would end up doing the same thing anyways.

—
You are receiving this because you commented.
Reply to this email directly or view it on GitHub
#3 (comment)

[giadefa]

let's talk about it when Noelia is in BCN

[j3mdamas]

@noeliaferruz, let's see this on monday.

[j3mdamas]

we fixed this when Noelia was here, but now there's only a general problem with patching that's getting solved (#97)

[stefdoerr]

So we can close this then. All issues are fixed for this as far as I can see (atom naming, backbone selections)

[j3mdamas]

yes, only issue 97 left.