Re: Reason: FATAL ERROR: DIDN'T FIND vdW PARAMETER FOR ATOM TYPE OT

From: Kenno Vanommeslaeghe (kvanomme_at_rx.umaryland.edu)
Date: Wed Sep 18 2013 - 13:50:04 CDT

Grace and Teerapong nailed it - you commented out the water parameter
file, therefore NAMD complains it doesn't find the L-J parameters for
water oxygen. Although this is pretty trivial, it is partially my fault
because I suggested to use the file as downloaded from our site, while I
should know that doesn't work with NAMD. The whole stroy is given below,
but allow me to first warn you to !!! NOT USE TEERAPONG'S FILE !!! He or
she commented out one NBFIX line too much, which will make any sodium ions
in your solvent stick UNPHYSICALLY strongly to any carboxylate groups in
your protein. Please read the explanation below. (I would also recommend
the NAMD DEVELOPERS TO READ IT because it seems we have an unsolved
problem on our hands that we didn't anticipate when releasing C36.) The
short story is that the best thing you can do right now is to read *all*
parameter files (even the ones you don't need) into NAMD, per the
instructions on my home page:
http://mackerell.umaryland.edu/~kenno/cgenff/program.html#namd
Referring to these instructions, you do still need to comment out
"parameters molecule.str", but nothing else.

The whole story
===============
There are 2 separate issues that prevent NAMD from reading
toppar_water_ions.str . One is the CHARMM commands, the other is the NBFIX
lines. Working around the former is relatively straightforward, but the
latter isn't.

The CHARMM commands
-------------------
These CHARMM commands were put into the file to work around a shortcoming
in the CHARMM parameter reader. While this shortcoming is presumably fixed
in the recent CHARMM 38 release (though I haven't confirmed this yet), the
workaround will presumably remain in the file a while longer to support
users of older CHARMM versions. As far as NAMD is concerned I see 2
possible workarounds:
(1) Release separate versions of all the stream files that don't contain
these commands
(2) Make the NAMD parameter reader ignore everything outside of the
parameter section (which starts with "read para" and ends with "end")

The NBFIX lines
---------------
These are actually pretty important, and cannot safely be removed. Their
function is to work around an inherent problem with the Class I additive
potential energy function (used in all major non-polarizable biomolecular
force fields, not only CHARMM). Specifically, the CHARMM sodium and
carboxylate nonbonded parameters were originally parameterized to interact
correctly with water (which is very important), but it was found shortly
afterwards that this *inherently* causes them to interact incorrectly with
each other. NBFIX stands for "nonbonded fix", and fixes the
sodium-carboxylate interaction. There are carboxylate groups with
different oxygen atom types in the protein, lipid, carbohydrate and
general force fields; that's why there are 4 NBFIX lines.

The big problem is that if you haven't read the lipid force field into
NAMD, it will choke on the lipid line; same thing for the other parts of
the CHARMM force field. The CHARMM parameter reader also has this issue,
but the BOMLEV and WRNLEV commands instruct it to ignore the unknown atom
types. As far as NAMD is concerned, I see 3 possible workarounds:
(1) Instruct all NAMD users to comment out *only* the lines associated
with the parts of the force fields they're *not* using. This workaround is
EXTREMELY DANGEROUS because even if all users manage to do the right thing
when editing the file, users within the same lab will inevitably use each
other's files for different simulations...
(2) Instruct all NAMD users to ALWAYS read all 4 carboxylate-containing
parameter files (protein, lipid, carbohydrate and general), even if such
molecules aren't present in the simulation system. While this not a very
clean thing to do, at least it's safe. Until a better solution is found, I
changed the NAMD instructions on my home page this way:
http://mackerell.umaryland.edu/~kenno/cgenff/program.html#namd
(3) Make the NAMD parameter reader not bail out when finding unrecognized
atom types in NBFIX lines (it would be even better to make it recognize
the BOMLEV and WRNLEV commands, but this may be a bit too nontrivial to
waste developer time on).

Sorry for the wall of text but I feel it is really important that we
(CHARMM and NAMD developers) find a solution to the NBFIX conundrum.

Cheers,

        Kenno.

On 09/18/2013 01:20 PM, Teerapong Pirojsirikul wrote:
> Hi,
>
> I saw you commented out toppar_water_ions.str which is required for
> the OT atom type. It seems to me that OT atom type is for TIP3P
> oxygen. My guess is you also used this file in psfgen and your system
> is solvated with TIP3P water, right?. However, I found it's a little
> bit tricky to also use this file in a namd configuratoin file. You
> need to delete or comment out a bunch of lines in it. I'd recommend
> trying my modified toppar_water_ions.mod.str attached. Just put it in
> your working directory and fix the line
>
> paraTypeCharmm on
> parameters ./par_all36_prot.prm
> parameters ./par_all36_na.prm
> #parameters ./toppar_water_ions.str
>
> to
>
> paraTypeCharmm on
> parameters ./par_all36_prot.prm
> parameters ./par_all36_na.prm
> parameters ./toppar_water_ions.mod.str
>
> ,and give me feedback.
>
> Cheers,
> Tee
>
>
> 2013/9/18 Anna Chailyan <anna.chailyan_at_gmail.com>:
>>
>> Hello Hello!
>>
>> I am desperate !
>>
>> Today i made a new psf file with topology topology top_all36_prot.rtf and
>> topology top_all36_na.rtf since my complex is made of protein and dna. It
>> runs fine and i obtain the psf file.
>> I attached the configuration file
>>
>> .
>> It keeps crashing when i run minimization by saying "FATAL ERROR: DIDN'T
>> FIND vdW PARAMETER FOR ATOM TYPE OT".
>>
>> I don't know, which are the possibilities.
>>
>>
>>
>> Best Regards,
>>
>>
>> On Sep 18, 2013, at 1:20 AM, Kenno Vanommeslaeghe
>> <kvanomme_at_rx.umaryland.edu> wrote:
>>
>> We recommend using the new CHARMM36 force field, which has significant
>> improvements, and has been made modular (ie. no more need to merge parameter
>> files). The force field can be downloaded from
>> http://mackerell.umaryland.edu/CHARMM_ff_params.html
>> Note that PSFs generated using the old force field usually won't work with
>> the new one; one has to feed psfgen the new topology files from the force
>> field distribution package. So assuming your system contains protein, lipids
>> and water, Niklaus' recommendations for psfgen would become:
>>
>> topology top_all36_prot.rtf
>> topology top_all36_lipid.rtf
>> topology toppar_water_ions.str # not sure it will accept this
>>
>> and for NAMD:
>>
>> paraTypeCharmm on
>> parameters par_all36_prot.prm
>> parameters par_all36_lipid.prm
>> parameters toppar_water_ions.str
>>
>> Cheers,
>>
>> Kenno.
>>
>> PS. For the record, the vintage *_all27_prot_lipids.* files were a merger of
>> the CHARMM22 + CMAP protein force field and the CHARMM27 lipid force field,
>> so the below methodology would have you read in obsolete CHARMM27 lipid
>> definitions and parameters, then overwrite them with CHARMM36 ones. While
>> I'm not sure this is a problem, it seems a bit risky.
>>
>>
>>
>> On 09/17/2013 12:35 PM, Niklaus Johner wrote:
>>
>> This looks like you didn't load the proper parameter files when running
>> your simulations.
>>
>> Typically when you generate your psf with psfgen (I guess?), you input
>> some topology files for example:
>>
>> topology top_all27_prot_lipids.rtf
>> topology top_all36_lipid.rtf
>>
>> So then in your namd input you need to load the cognate parameter files.
>> Something like:
>>
>> paraTypeCharmm on
>> parameters par_all27_prot_lipids.prm
>> parameters par_all36_lipid.prm
>>
>> Otherwise some of the atoms defined in your psf file won't be defined in
>> the parameter files.
>>
>> N.
>>
>>
>> Niklaus Johner
>> Weill Cornell Medical College
>> Harel Weinstein Lab
>> Department of Physiology and Biophysics
>> 1300 York Avenue, Room D-501
>> New York, NY 10065
>>
>>
>>
>> On Sep 17, 2013, at 12:10 PM, Anna Chailyan wrote:
>>
>>
>> Help!
>>
>> After all the previous successful steps finally i have psf file of my
>> complex and now when i try to minimize it i get this message:
>>
>>
>>
>> Reason: FATAL ERROR: DIDN'T FIND vdW PARAMETER FOR ATOM TYPE OT
>>
>> Charm++ fatal error:
>> FATAL ERROR: DIDN'T FIND vdW PARAMETER FOR ATOM TYPE OT
>>
>>
>> I am using NAMD_2.9_MacOSX-x86_64-multicore and i wonder why do i get
>> this message now, not when i actually generated psf file.
>> Please ! if there is someone that got this problem and knows how to
>> solve it, share it with me!
>> Best regards,
>> Anna
>>
>>
>>
>>
>>

This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:21:41 CST