From: Jérôme Hénin (jerome.henin_at_ibpc.fr)
Date: Tue Jul 11 2017 - 11:52:23 CDT
Yup, the conditions for stable dynamics should not be different in the ABF
case than in any MD simulation.
Jerome
On 11 July 2017 at 18:28, Francesco Pietra <chiendarret_at_gmail.com> wrote:
> Hi Jerome:
> With a new definition of colvars, replacing the absurd one I had first
> devised, ABF Conformation Bound runs OK. Thank you very much.
>
> A side question: I am using "rigid bonds water" "ts=1.0fs". Is that safe
> rigidifying all bonds, to speed up, with ts=2.0fs, in these ABF and FEP
> simulations?
>
> thanks
> francesco
>
> On Tue, Jul 11, 2017 at 3:12 PM, Jérôme Hénin <jerome.henin_at_ibpc.fr>
> wrote:
>
>> Hi Francesco,
>>
>> I thought your messages were self-explanatory. This looks like a mistake
>> in the colvar definition.
>>
>> I don't think you need to enable forceNoPBC, by the way.
>>
>> Jerome
>>
>> On 11 July 2017 at 08:20, Francesco Pietra <chiendarret_at_gmail.com> wrote:
>>
>>> hello jerome:
>>> Did you see the r colvar that I posted, and my doubt, now, whether "r"
>>> should refer to the distance between the centers of mass of ligand and
>>> protein or rather to the distance between the ligand and the protein
>>> residues facing the ligand? What I get from "cv printframe" is a short
>>> distance.
>>>
>>> thanks
>>> francesco
>>> ---------- Forwarded message ----------
>>> From: Francesco Pietra <chiendarret_at_gmail.com>
>>> Date: Sun, Jul 9, 2017 at 7:44 AM
>>> Subject: Re: namd-l: ABF FEP with multichain receptor
>>> To: Jérôme Hénin <jerome.henin_at_ibpc.fr>
>>> Cc: Namd Mailing List <namd-l_at_ks.uiuc.edu>
>>>
>>>
>>> Colvarstrajfrequency 100
>>> Colvarsrestartfrequency 100
>>>
>>>
>>> #############################################################
>>>>> # rmsd PMF
>>>>> # r, Theta, Phi, Psi, theta, phi UNRESTRAINED
>>>>> # r and polar angles theta phi defining the position of ligand with
>>>>> respect to protein
>>>>> # Euler angles Theta Phi Psi define ligand orientation
>>>>> #############################################################
>>>>> # COLVARS DEFINITIONS
>>>>> #############################################################
>>>>>
>>>>>
>>>>> colvar {
>>>>> name r
>>>>>
>>>>> width 1.0
>>>>>
>>>>> lowerboundary 0.0
>>>>> upperboundary 40.0
>>>>>
>>>>> lowerwallconstant 100.0
>>>>> upperwallconstant 100.0
>>>>>
>>>>> distance {
>>>>> forceNoPBC yes
>>>>> group1 {
>>>>> atomnumbers : { 12258 12259 12260 12261 12280 12281 } # PROT
>>>>> R359: N HN CA HA C O
>>>>> }
>>>>> group2 {
>>>>> atomnumbers { 13439 13438 13437 13445 13446 13448 } # SAA1
>>>>> hydrofuran: C2 C1 O1 C8 C9 C11
>>>>> }
>>>>> }
>>>>> }
>>>>>
>>>>
>>> ARG 359 faces the ligand atoms in group 2
>>>
>>> On Sat, Jul 8, 2017 at 10:07 PM, Jérôme Hénin <jerome.henin_at_ibpc.fr>
>>> wrote:
>>>
>>>> Can you send the part of your colvars input that defines the distance
>>>> colvar?
>>>>
>>>> Jerome
>>>>
>>>> On 8 July 2017 at 18:08, Francesco Pietra <chiendarret_at_gmail.com>
>>>> wrote:
>>>>
>>>>> Hi Jerome:
>>>>> That helps, thank you.
>>>>>
>>>>> My question stemmed from observing a major discrepancy for the
>>>>> distance between the centers of mass of the protein and the ligand
>>>>> according to how it is measured.
>>>>>
>>>>> With the same files from a "wrapAll no" MD simulation (just now
>>>>> completed) loaded to vmd,
>>>>>
>>>>> set selprot [atomselect top "protein"]
>>>>> measure center $selprot
>>>>> set sellig [atomselect top "segname SAA1"]
>>>>> measure center $sellig
>>>>> set dist [ veclength [vecsub [measure center $selprot] [measure
>>>>> center $sellig]]]
>>>>>
>>>>> gives 28A
>>>>>
>>>>> while cv printframe gives 8A, which is totally unrealistic, much too
>>>>> small.
>>>>>
>>>>> This is just the same that I observed from MD simulations with
>>>>> "wrapAll on".
>>>>>
>>>>> My first hypothesis was that "cv printframe", for reasons that I do
>>>>> not understand, is taking into account the sole chain of the receptor where
>>>>> the ligand resides (although even so a 8A is probably not enough. As "cv
>>>>> printframe" is wrong about "r", perhaps it is also wrong about Theta, etc.
>>>>>
>>>>> I assume that I am doing something wrong and I am embarrassed in
>>>>> coming again here with the same problem (however, I am now here with a MD
>>>>> "wrapAll no", as required for measuring centers when there is more than one
>>>>> molecule. I do not see any overlapping between the periodic boxes, and the
>>>>> whole protein-ligand is within. There are two other ligands, one in each
>>>>> chain, common ligands in biochemistry.
>>>>>
>>>>> Thanks for your (and all guys) understanding.
>>>>>
>>>>> francesco pietra
>>>>>
>>>>> On Sat, Jul 8, 2017 at 12:46 PM, Jérôme Hénin <jerome.henin_at_ibpc.fr>
>>>>> wrote:
>>>>>
>>>>>> Hi Francesco,
>>>>>>
>>>>>> That should not be a problem, as long that you make sure your
>>>>>> multi-chain receptor always stays in one piece, and is never split across
>>>>>> several periodic cells
>>>>>>
>>>>>> Jerome
>>>>>>
>>>>>> On 7 July 2017 at 17:21, Francesco Pietra <chiendarret_at_gmail.com>
>>>>>> wrote:
>>>>>>
>>>>>>> Hello:
>>>>>>> Does anyone know whether ABF and FEP simulations can be carried out
>>>>>>> with namd on a two-chains protein containing a ligand in one of the two
>>>>>>> chains?
>>>>>>> i am referring to, in particular, the 2017 tutorial protein-ligand.
>>>>>>>
>>>>>>> thanks
>>>>>>>
>>>>>>> francesco pietra
>>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>>
>>
>
This archive was generated by hypermail 2.1.6 : Mon Dec 31 2018 - 23:20:26 CST