From: Giacomo Fiorin (giacomo.fiorin_at_gmail.com)
Date: Wed Nov 17 2021 - 13:42:14 CST
Please stop changing your parameters at random: it won't help you either in
the short or long term.
The best place would be to pick the first of those PDB structures, and
compute the RMSD of the chosen atoms in the initial configuration for the
NAMD run and the one from the PDB file, e.g. "1.pdb". This can be done by
setting a simple rmsd function instead of the path CV:
https://urldefense.com/v3/__https://colvars.github.io/colvars-refman-namd/colvars-refman-namd.html*sec:cvc_rmsd__;Iw!!DZ3fjg!vKPC7M13JpfoITkSSfCNsSRs4_ziQtMYbpAAWkgd4La7ChXIFTVJ0p_BEVnaxWFvLQ$
Note that the PDB files may contain the whole system or just the selection:
https://urldefense.com/v3/__https://colvars.github.io/colvars-refman-namd/colvars-refman-namd.html*sec:colvars_pdb_format__;Iw!!DZ3fjg!vKPC7M13JpfoITkSSfCNsSRs4_ziQtMYbpAAWkgd4La7ChXIFTVJ0p_BEVm426Z0XQ$
but in other contexts (e.g. in VMD) you can only load a PDB file that
matches the same molecule. Take this into account.
If everything is set up correctly and an optimal rotation is possible, you
should be able to compute the same RMSDs using VMD's internal commands and
Colvars. When computed over atoms of the same mass, there is an identical
match between "measure rmsd" in VMD and the "rmsd" function in Colvars. If
they are not the same, something is different in your configuration.
Note that you can also use the Colvars Dashboard, included as a plugin in
the latest VMD alpha release:
https://urldefense.com/v3/__https://colvars.github.io/colvars-refman-vmd/colvars-refman-vmd.html*sec:dashboard__;Iw!!DZ3fjg!vKPC7M13JpfoITkSSfCNsSRs4_ziQtMYbpAAWkgd4La7ChXIFTVJ0p_BEVlXCF6u_w$
to avoid having to set up a new NAMD run while you're debugging.
Once you see that the RMSD can be computed reproducibly between the initial
simulation snapshot and the first frame, move over to the other PNEB frames.
Giacomo
On Wed, Nov 17, 2021 at 2:12 PM Ashutosh Shandilya <scfbioiitd_at_gmail.com>
wrote:
> I increased harmonic restrain to 5.0 using command
> abf {
> colvars gs
> fullSamples 1000
> historyfreq 100000
> hidejacobian off
> }
> harmonic {
> colvars gs
> centers 0.0
> forceconstant 5.0
> }
>
> Output file shows the same RMSD for all the structures something like this
> which should not be the case. Is it because i switched off
> *usesecondclosestframe* off. I have been trying to change several
> parameters in the input but still getting some kind of error.
>
> Frame index: 67 ; optimal RMSD = 0.985712
> Frame index: 80 ; optimal RMSD = 0.985712
> Frame index: 79 ; optimal RMSD = 0.985712
> Frame index: 78 ; optimal RMSD = 0.985712
> Frame index: 77 ; optimal RMSD = 0.985712
> Frame index: 76 ; optimal RMSD = 0.985712
> Frame index: 75 ; optimal RMSD = 0.985712
> Frame index: 74 ; optimal RMSD = 0.985712
> Frame index: 73 ; optimal RMSD = 0.985712
> Frame index: 72 ; optimal RMSD = 0.985712
> Frame index: 71 ; optimal RMSD = 0.985712
> Frame index: 70 ; optimal RMSD = 0.985712
> Frame index: 69 ; optimal RMSD = 0.985712
> Frame index: 68 ; optimal RMSD = 0.985712
> Frame index: 81 ; optimal RMSD = 0.985712
> Frame index: 66 ; optimal RMSD = 0.985712
> Frame index: 65 ; optimal RMSD = 0.985712
> Frame index: 64 ; optimal RMSD = 0.985712
> Frame index: 63 ; optimal RMSD = 0.985712
> Frame index: 62 ; optimal RMSD = 0.985712
> Frame index: 61 ; optimal RMSD = 0.985712
> Frame index: 60 ; optimal RMSD = 0.985712
> Frame index: 59 ; optimal RMSD = 0.985712
> Frame index: 58 ; optimal RMSD = 0.985712
> Frame index: 57 ; optimal RMSD = 0.985712
> Frame index: 56 ; optimal RMSD = 0.985712
> Frame index: 55 ; optimal RMSD = 0.985712
> Frame index: 94 ; optimal RMSD = 0.985712
> Frame index: 107 ; optimal RMSD = 0.985712
>
>
> 108 PDB structures were obtained from the PNEB path. I computed RMSD with
> VMD which is mentioned below. Kindly have a look and suggest. My actual
> RMSD varies like this.
> 1 0.5520958304405212
> 2 0.6408998370170593
> 3 0.7004873752593994
> 4 0.7478375434875488
> 5 0.7849769592285156
> 6 0.8124300241470337
> 7 0.829501748085022
> 8 0.8617943525314331
> 9 0.8889437317848206
> 10 0.9140389561653137
> 11 0.925011396408081
> 12 0.931750476360321
> 13 0.9490005373954773
> 14 0.9639561176300049
> 15 0.9865623712539673
> 16 1.0129530429840088
> 17 1.017659068107605
> 18 1.0304405689239502
> 19 1.05254328250885
> 20 1.0523943901062012
> 21 1.0497472286224365
> 22 1.0674532651901245
> 23 1.0869779586791992
> 24 1.1080254316329956
> 25 1.1066851615905762
> 26 1.1148474216461182
> 27 1.1117987632751465
> 28 1.1251643896102905
> 29 1.1315655708312988
> 30 1.1351579427719116
> 31 1.1500359773635864
> 32 1.1598128080368042
> 33 1.1690787076950073
> 34 1.1832456588745117
> 35 1.190578579902649
> 36 1.1949034929275513
> 37 1.2126442193984985
> 38 1.2073010206222534
> 39 1.2080799341201782
> 40 1.2127560377120972
> 41 1.2220537662506104
> 42 1.2297075986862183
> 43 1.2463030815124512
> 44 1.258796215057373
> 45 1.2574396133422852
> 46 1.2649439573287964
>
> Kindly suggest.
>
> Thanks,
> Ashutosh
>
>
> On Wed, Nov 17, 2021 at 1:32 PM Giacomo Fiorin <giacomo.fiorin_at_gmail.com>
> wrote:
>
>> There *were* multiple RMSD computations, because that's how the
>> distances between path frames are computed.
>>
>> You are also omitting several warnings above, btw. One was "Please check
>> your configuration or increase restraint on z(σ)". What did you do there?
>>
>> I don't know where you got the PDB structures from, but I recommend that
>> you load them in VMD and compute the RMSD of the coordinates of your
>> simulation with respect to each. All RMSDs need to be well defined in
>> order to be able to compute the path variable derived from them.
>>
>> Giacomo
>>
>>
>> On Wed, Nov 17, 2021 at 12:55 PM Ashutosh Shandilya <scfbioiitd_at_gmail.com>
>> wrote:
>>
>>> Thanks for your reply, As I understand I should consider the selection
>>> which has only moderate change. First step itself gives this problem so I
>>> have not calculated the RMSD. I have a string of pdbs from PNEB path and I
>>> want to have a pmf that's why I am using abf. So the selected atoms have
>>> lots of transformation. So what are the options.
>>>
>>> Many thanks,
>>> Ashutosh
>>>
>>> On Wed, Nov 17, 2021 at 11:28 AM Giacomo Fiorin <
>>> giacomo.fiorin_at_gmail.com> wrote:
>>>
>>>> This message is the key:
>>>> colvars: This is usually the result of an ill-defined set of atoms for
>>>> rotational alignment (RMSD, rotateReference, etc).
>>>>
>>>> If the structure of the group of atoms changes too much, it becomes
>>>> possible that a pure rotation is not the transformation that minimizes the
>>>> RMSD, but another transformation does (e.g. rotation + reflection across a
>>>> plane).
>>>>
>>>> You do not need to change the time step, but instead pick a selection
>>>> that is guaranteed to have only moderate changes in structure. Practically
>>>> speaking, the RMSD should remain much lower than the radius of gyration.
>>>>
>>>> If you compute that RMSD in VMD you will have a similar problem, just
>>>> with a different outcome because the numerical algorithm is different.
>>>>
>>>> Giacomo
>>>>
>>>> On Wed, Nov 17, 2021 at 11:03 AM Ashutosh Shandilya <
>>>> scfbioiitd_at_gmail.com> wrote:
>>>>
>>>>> I am running ABF and getting this error.
>>>>>
>>>>> Frame index: 32 ; optimal RMSD = 4.87456
>>>>> Frame index: 31 ; optimal RMSD = 4.87456
>>>>> Frame index: 30 ; optimal RMSD = 4.87456
>>>>> Frame index: 29 ; optimal RMSD = 4.87456
>>>>> Frame index: 28 ; optimal RMSD = 4.87456
>>>>> colvars: Prepared sample and gradient buffers at step 0.
>>>>> colvars: Too many iterations in routine jacobi.
>>>>> colvars: This is usually the result of an ill-defined set of atoms for
>>>>> rotational alignment (RMSD, rotateReference, etc).
>>>>> colvars: Too many PFATAL ERROR: Error in the collective variables
>>>>> module (see above for details)
>>>>> FATAL ERROR: Error in the collective variables module (see above for
>>>>> details)
>>>>>
>>>>> I have used different timesteps but still I am getting this error. Any
>>>>> suggestions?
>>>>>
>>>>> Many thanks
>>>>>
>>>>
This archive was generated by hypermail 2.1.6 : Fri Dec 31 2021 - 23:17:12 CST