From: Marcelo C. R. Melo (melomcr_at_gmail.com)
Date: Sat Apr 04 2020 - 11:25:27 CDT
Hi Francesco,
I'm glad to be of help, but without a minimal test case it is hard to help
you further.
Given that the issue is so inconsistent, you should provide a minimal set
of file that reproduce the issue, in the same NAMD version.
This would also help other who are trying to set up parameters for their
QM/MM simulations.
Best,
Marcelo
On Fri, 3 Apr 2020 at 12:49, Francesco Pietra <chiendarret_at_gmail.com> wrote:
> Hi Marcelo, Hi Josh
> I set aside for the moment the above problem, going on with other ligands
> of the same nucleosides. The new ligands, as before, only differ for a
> substituent at the rings. I could run QMMM without problems for another two
> systems, the third one gave again the problem of params:
>
> Info: SUMMARY OF PARAMETERS:
>> Info: 810 BONDS
>> Info: 2329 ANGLES
>> Info: 5369 DIHEDRAL
>> Info: 176 IMPROPER
>> Info: 6 CROSSTERM
>> Info: 321 VDW
>> Info: 2 VDW_PAIRS
>> Info: 0 NBTHOLE_PAIRS
>> FATAL ERROR: DIDN'T FIND vdW PARAMETER FOR ATOM TYPE CG2R_1
>>
>
> this time vdW, not angle.
>
> Notice that NAMD had no problems in running MD with this system and same
> .str parameter set.
>
> Notice also that at the above crash point, NAMD had already read other
> params, including those for the ligand. I can say that from warnings like
>
> Warning: DUPLICATE vdW ENTRY FOR CG2R51
>> PREVIOUS VALUES sigma=3.54578 epsilon=0.073 sigma14=3.54578
>> epsilon14=0.073
>> USING VALUES sigma=3.74177 epsilon=0.05 sigma14=3.74177 epsilon14=0.05
>>
>
> which relate to the ligand, and are the same as for the "good" cases
>
> Notice, finally, that the .str parameter set has exactly the same layout
> in all cases, the good ones and the bad ones.
>
> Trying to pave the way to NAMD, I separated the .str into topology and
> parameters, using the latter, Same issue.
>
> What NAMD was looking for was present in .str and .prm in the form:
>
> HT_W 0.000000 -0.046000 0.224500 ! TIP3P HYDROGEN PARAMETERS, adm
>> jr., NBFIX obsolete
>> OT_W 0.000000 -0.152100 1.768200 ! TIP3P OXYGEN PARAMETERS, adm
>> jr., NBFIX obsolete
>>
>> CG3R_0 0.00 -0.0320 2.0000
>> CG2D_0 0.00 -0.0680 2.0900
>> CG2D_1 0.00 -0.0680 2.0900
>> CG2R_0 0.00 -0.0730 1.9900
>> CG2R_1 0.00 -0.0730 1.9900
>> CG32_0 0.00 -0.0560 2.0100
>>
>> END
>>
>
>
> So, in my hands, NAMD_Git-2020-02-24_Linux-x86_64-multicore, either on my
> home at the cluster or on my personal linux box, has scattered problems in
> reading parameters when running QMMM, not when running MD.
>
> Last year I carried out at the same cluster a large number of QMMM
> simulations, using the same vmd ... prepare, for quite different systems
> and with a previous version of NAMD (can't say which one, usually I use,
> like now, nightbuilds). I never met the above problems of params.
>
> The problem is so curious and baffling that I decided to set aside, from
> now, the "bad" cases, working on the "good" ones (in all "good" cases
> convergence gave no problems at a high DFT level). Perhaps I am missing
> something. If so, it will emerge soon or later on rethinking. For now,
> there is no point to change NAMD version.
>
> Thanks for the kind advice that I received. I'll be back with new facts or
> ideas on this problem of params, hopefully soon.
>
> francesco
>
>
>
>
> On Thu, Apr 2, 2020 at 6:41 PM Josh Vermaas <joshua.vermaas_at_gmail.com>
> wrote:
>
>> Hi Francesco,
>>
>> If NAMD can't find parameters, it'll bail. Looking back through the email
>> chain, where in the parameter files that you passed *should* NAMD find the
>> CG331 NG2R51 CG3RC1 parameters? They look like CGenFF atomtypes. Did you
>> perhaps forget to pass the molecule to CGenFF or something else to provide
>> dummy inputs? To be honest, if you are sure that the only place this angle
>> occurs is in the QM region, I think its safe to just add a line to your
>> parameter files like this to provide NAMD with something it can put in its
>> memory banks:
>>
>> CG331 NG2R51 CG3RC1 0 0
>>
>> -Josh
>>
>> On Thu, Apr 2, 2020 at 3:20 AM Francesco Pietra <chiendarret_at_gmail.com>
>> wrote:
>>
>>> Hi Marcelo
>>> Although I understand that NAMD first checks the parameters, for the
>>> system at issue I tried now QMMM with a smaller set of QM atoms around
>>> ligand YYY. This includes only the nucleosides that were seen by X-ray
>>> diffr to interact with a closely similar small ligand XXX, where no issue
>>> arose. Five nucleosides, thus same charge -5, and no unpaired electron, in
>>> both cases.
>>> Same issue with YYY, the simulation crashing at
>>>
>>> Info: SUMMARY OF PARAMETERS:
>>>> Info: 810 BONDS
>>>> Info: 2329 ANGLES
>>>> Info: 5369 DIHEDRAL
>>>> Info: 176 IMPROPER
>>>> Info: 6 CROSSTERM
>>>> Info: 321 VDW
>>>> Info: 2 VDW_PAIRS
>>>> Info: 0 NBTHOLE_PAIRS
>>>> FATAL ERROR: UNABLE TO FIND ANGLE PARAMETERS FOR CG331 NG2R51 CG3RC1
>>>> (ATOMS 3958 3967 3968)
>>>> [Partition 0][Node 0] End of program
>>>>
>>>
>>> in contrast with the simulation with ligand XXX
>>>
>>> Info: SUMMARY OF PARAMETERS:
>>>> Info: 804 BONDS
>>>> Info: 2310 ANGLES
>>>> Info: 5320 DIHEDRAL
>>>> Info: 176 IMPROPER
>>>> Info: 6 CROSSTERM
>>>> Info: 320 VDW
>>>> Info: 2 VDW_PAIRS
>>>> Info: 0 NBTHOLE_PAIRS
>>>> Info: TIME FOR READING PSF FILE: 0.435593
>>>> Info: Reading pdb file heat_restr50-01.pdb
>>>> Info: TIME FOR READING PDB FILE: 0.0747963
>>>> Info:
>>>> Info: Using the following PDB file for QM parameters: sphere20-qm.pdb
>>>> Info: Number of QM atoms (excluding Dummy atoms): 190
>>>> Info: We found 8 QM-MM bonds.
>>>> Info: Applying user defined multiplicity 1 to QM group ID 1
>>>> Info: 1) Group ID: 1 ; Group size: 190 atoms ; Total PSF charge: -5
>>>>
>>>
>>> Does this suggest anything or is it just what should happen when NAMD
>>> has problems with params?
>>>
>>> All the best
>>> francesco
>>>
>>>
>>>
>>> On Thu, Apr 2, 2020 at 12:05 AM Marcelo C. R. Melo <melomcr_at_gmail.com>
>>> wrote:
>>>
>>>> Hi Francesco,
>>>>
>>>> You did a thorough job testing this, so I'm running out of ideas.
>>>> Last thing that comes to mind is double checking the definition o
>>>> the QM region. NAMD removes the bonded interactions between QM atoms (to
>>>> avoid double counting the interactions already covered by the QM software),
>>>> but this happens *after* NAMD checks for parameters. Still, maybe there is
>>>> something being left out?
>>>> Let me know how this goes.
>>>> Best,
>>>> Marcelo
>>>>
>>>> On Wed, 1 Apr 2020 at 17:16, Francesco Pietra <chiendarret_at_gmail.com>
>>>> wrote:
>>>>
>>>>> Hi Marcelo
>>>>> It is more than a puzzle for me.
>>>>>
>>>>> I restarted from the folder where minimization, heating, and NVT had
>>>>> been run for this system.
>>>>>
>>>>> I checked that NVT could be started again.
>>>>>
>>>>> Then, I built the files for qmmm (vmd ....prepare) into this folder
>>>>> with just the files that had served for NVT.
>>>>>
>>>>> Well, namd-01.conf (qmmm) gave the same error that I reported, while
>>>>> nvt-01.conf (classical MD) could be run again without problems.
>>>>>
>>>>> It is implied (but was also checked, and there are no voids in the
>>>>> .str) that the angle at issue is correctly defined.
>>>>>
>>>>> Yes, I had no problems with both MD and QMMM with other four very
>>>>> similar system, where the small molecule differs for only a substituent at
>>>>> a ring. Owing to that, I hope that I'll be able to think better tomorrow.
>>>>>
>>>>> All the best
>>>>> francesco
>>>>>
>>>>>
>>>>>
>>>>> On Wed, Apr 1, 2020 at 8:16 PM Marcelo C. R. Melo <melomcr_at_gmail.com>
>>>>> wrote:
>>>>>
>>>>>> Hi Francesco,
>>>>>>
>>>>>> It is curious that you have similar systems running without issue.
>>>>>> Can I guess that you created the parameters using CGenFF (or similar)?
>>>>>>
>>>>>> Maybe something in the topology definitions of this molecule is
>>>>>> creating an angle that doesn't have a parameter. Should this angle be
>>>>>> defined at all? It is hard to discuss without knowing more about the
>>>>>> system, but maybe something in the automatic topology creation got messed
>>>>>> up.
>>>>>>
>>>>>> Best,
>>>>>> Marcelo
>>>>>>
>>>>>> On Wed, 1 Apr 2020 at 13:46, Francesco Pietra <chiendarret_at_gmail.com>
>>>>>> wrote:
>>>>>>
>>>>>>> I Marcelo
>>>>>>> I assumed that, providing all files from prepare, it will run, as it
>>>>>>> was the case on quite similar systems.
>>>>>>>
>>>>>>> The cluster was busy and it has only now tried to run the simulation
>>>>>>> with the same error message.
>>>>>>> Following your suggestion "some naming conflict for the atoms you
>>>>>>> mentioned (maybe double check the PSF file and the str file?)."
>>>>>>> I was unable to find any conflict for the atoms I mentioned. Here
>>>>>>> excerpts:
>>>>>>>
>>>>>>> .str file
>>>>>>> ATOM C5 CG3RC1 0.212
>>>>>>> ATOM C6 CG3C52 -0.073
>>>>>>> ATOM C7 CG3RC1 -0.138
>>>>>>> ATOM C8 CG3RC1 0.092
>>>>>>> ATOM C9 CG2D_0 0.434
>>>>>>> ATOM C10 CG2D_1 0.012
>>>>>>> ATOM C11 CG2R_0 -0.202
>>>>>>> ATOM C12 CG2R51 -0.235
>>>>>>> ATOM C13 CG2R51 -0.057
>>>>>>> ATOM N NG2R51 -0.021
>>>>>>> ATOM C15 CG331 -0.188
>>>>>>>
>>>>>>> *QMonly.psf file
>>>>>>> 324 MO4 640 ACL C5 CG3RC1 0.212000
>>>>>>> 12.0110 0
>>>>>>> 325 MO4 640 ACL C6 CG3C52 -0.073000
>>>>>>> 12.0110 0
>>>>>>> 326 MO4 640 ACL C7 CG3RC1 -0.138000
>>>>>>> 12.0110 0
>>>>>>> 327 MO4 640 ACL C8 CG3RC1 0.092000
>>>>>>> 12.0110 0
>>>>>>> 328 MO4 640 ACL C9 CG2D_0 0.434000
>>>>>>> 12.0110 0
>>>>>>> 329 MO4 640 ACL C10 CG2D_1 0.012000
>>>>>>> 12.0110 0
>>>>>>> 330 MO4 640 ACL C11 CG2R_0 -0.202000
>>>>>>> 12.0110 0
>>>>>>> 331 MO4 640 ACL C12 CG2R51 -0.235000
>>>>>>> 12.0110 0
>>>>>>> 332 MO4 640 ACL C13 CG2R51 -0.057000
>>>>>>> 12.0110 0
>>>>>>> 333 MO4 640 ACL N NG2R51 -0.021000
>>>>>>> 14.0070 0
>>>>>>> 334 MO4 640 ACL C15 CG331 -0.188000
>>>>>>> 12.0110
>>>>>>>
>>>>>>> general .psf file
>>>>>>> 3958 MO4 640 ACL C5 CG3RC1 0.212000
>>>>>>> 12.0110 0
>>>>>>> 3959 MO4 640 ACL C6 CG3C52 -0.073000
>>>>>>> 12.0110 0
>>>>>>> 3960 MO4 640 ACL C7 CG3RC1 -0.138000
>>>>>>> 12.0110 0
>>>>>>> 3961 MO4 640 ACL C8 CG3RC1 0.092000
>>>>>>> 12.0110 0
>>>>>>> 3962 MO4 640 ACL C9 CG2D_0 0.434000
>>>>>>> 12.0110 0
>>>>>>> 3963 MO4 640 ACL C10 CG2D_1 0.012000
>>>>>>> 12.0110 0
>>>>>>> 3964 MO4 640 ACL C11 CG2R_0 -0.202000
>>>>>>> 12.0110 0
>>>>>>> 3965 MO4 640 ACL C12 CG2R51 -0.235000
>>>>>>> 12.0110 0
>>>>>>> 3966 MO4 640 ACL C13 CG2R51 -0.057000
>>>>>>> 12.0110 0
>>>>>>> 3967 MO4 640 ACL N NG2R51 -0.021000
>>>>>>> 14.0070 0
>>>>>>> 3968 MO4 640 ACL C15 CG331 -0.188000
>>>>>>> 12.0110
>>>>>>>
>>>>>>> This isolated failure is so surprising to me that I would like to
>>>>>>> clear the matter. Do you have any other suggestion where i could check?
>>>>>>>
>>>>>>> thanks again
>>>>>>> francesco
>>>>>>>
>>>>>>>
>>>>>>> ---------- Forwarded message ---------
>>>>>>> From: Francesco Pietra <chiendarret_at_gmail.com>
>>>>>>> Date: Wed, Apr 1, 2020 at 6:09 PM
>>>>>>> Subject: Re: namd-l: QMMM param issue
>>>>>>> To: Marcelo C. R. Melo <melomcr_at_gmail.com>
>>>>>>> Cc: NAMD <namd-l_at_ks.uiuc.edu>
>>>>>>>
>>>>>>>
>>>>>>> Hi Marcelo
>>>>>>> I should have transcribed the original
>>>>>>>
>>>>>>> UNABLE TO FIND ANGLE PARAMETERS FOR CG331 NG2R51 CG3RC1 (ATOMS 3958
>>>>>>> C5,
>>>>>>> 3967 N, 3968 C15)
>>>>>>>
>>>>>>> Actually, that was a false message of error. The .str file was at
>>>>>>> the cluster, while the problem arose from not having sent to the cluster
>>>>>>> all files from the qmmm preparation (vmd -dispdev text -e prepare..tcl) on
>>>>>>> my laptop.
>>>>>>>
>>>>>>> Sorry for my lack of care. I should have better waited before
>>>>>>> posting, I missed to follow the rule that one should rethink several times.
>>>>>>>
>>>>>>> All the best
>>>>>>> francesco
>>>>>>>
>>>>>>>
>>>>>>> On Wed, Apr 1, 2020 at 4:58 PM Marcelo C. R. Melo <melomcr_at_gmail.com>
>>>>>>> wrote:
>>>>>>>
>>>>>>>> Hi Francesco,
>>>>>>>>
>>>>>>>> NAMD check for available parameters for every system you try to
>>>>>>>> load (QMMM or not). Even if you have a QM region with no parameters, which
>>>>>>>> will never be simulated classically, you still need to provide NAMD with
>>>>>>>> "dummy" parameters to allow the simulation to start (QwikMD can help you
>>>>>>>> create dummy parameters for QM regions).
>>>>>>>>
>>>>>>>> Therefore, this is likely an input problem. Either your NAMD config
>>>>>>>> file is not sourcing the right parameter file (*.str), or your str file has
>>>>>>>> some naming conflict for the atoms you mentioned (maybe double check the
>>>>>>>> PSF file and the str file?).
>>>>>>>>
>>>>>>>> In fact, the names of the atom types in your str file suggest three
>>>>>>>> carbon atoms (CG3RC1 CG3C52 CG3RC1 ), while in your explanation you
>>>>>>>> mention two carbon and one nitrogen atoms (C5, N, C15), with different atom
>>>>>>>> types, but maybe that was a typo on your part (?).
>>>>>>>>
>>>>>>>> Best,
>>>>>>>> Marcelo
>>>>>>>>
>>>>>>>> On Wed, 1 Apr 2020 at 07:28, Francesco Pietra <
>>>>>>>> chiendarret_at_gmail.com> wrote:
>>>>>>>>
>>>>>>>>> Hello
>>>>>>>>> With a QMMM (NAMD_Git-2020-02-24_Linux-x86_64-multicore on my home
>>>>>>>>> on the cluster) for a nucleic-smallmolecule system in TIP3, immediate
>>>>>>>>> crashwith error
>>>>>>>>>
>>>>>>>>> UNABLE TO FIND ANGLE PARAMETERS FOR CG331 NG2R51 CG3RC1 (ATOMS
>>>>>>>>> 3958 C5, 3967 N, 3968 C15) [atom names added by myself]
>>>>>>>>>
>>>>>>>>> Actually, params for the requested angle are present in the .str
>>>>>>>>> file
>>>>>>>>> CG3RC1 CG3C52 CG3RC1 58.00 105.30
>>>>>>>>>
>>>>>>>>> This issue did not arise with minimizations and heating for the
>>>>>>>>> above system. I can also add that QMMM for a variety of similar systems,
>>>>>>>>> differing only slightly (and not for the atoms at issue) in the
>>>>>>>>> "smallmolecule" went on without any issue.
>>>>>>>>>
>>>>>>>>> The issue was reproduced on launching the simulation at the
>>>>>>>>> cluster from scratch and I fall short of realizing from where the issue can
>>>>>>>>> arise.
>>>>>>>>> Thanks for suggestions
>>>>>>>>>
>>>>>>>>> francesco pietra
>>>>>>>>>
>>>>>>>>
This archive was generated by hypermail 2.1.6 : Fri Dec 31 2021 - 23:17:08 CST