[gmx-users] Re: Gromacs QM-MM Tutorial

[Praveen Agrawal]

Hi,
I have finally got the gromacs+mopac working, but on running it gives the
error:

QM/MM calculation requested.
there we go!
Layer 0
nr of QM atoms 24
QMlevel: PM3/6-31G*

 THERE ARE 51 DOUBLY FILLED LEVELS


          RHF CALCULATION, NO. OF DOUBLY OCCUPIED LEVELS = 51

                           NO. OF SINGLY OCCUPIED LEVELS =  1

 KEYWORD 'ANALYT' CANNOT BE USED HERE:
 ANALYICAL C.I. DERIVATIVES MUST USE FINITE DIFFERENCES
 TO CORRECT, REMOVE KEYWORD 'ANALYT' OR ADD 'NOANCI'


None of the input says 'analyt', Can anyone suggest as to what is wrong?

Best Regards,
Praveen

>
> Hi Dr.Groenhof,
> Thanks for the reply. I started working on that and than realized don't
> have MOPAC plugged up with GROMACS. It took a while to get it running,
> your installation comments were really helpful. I am below giving the
> exact route to get the things working (little different from your tutorial
> because of change in compiler version), might be of help to others.
>
> Best Regards,
> Praveen
>
> Machine pentium dual core with Linux 2.6.24-23-generic 32 bit (i686)
>
> 1. downloaded gromacs-4.0.4 & libmopac.a from site
> http://wwwuser.gwdg.de/~ggroenh/qmmm.html
> 2. libmopac.a goes to /usr/local/lib
> 3. Created test.c with
>
>      #include <ctype.h>
>
>              __const unsigned short int **__wrap___ctype_b (void) {
>                return __ctype_b_loc();
>              }
>
> 4. cc -c test.c to creat test.o
> 5. ar rcv libtest.a test.o; ranlib libtest.a
> 6. libtest.a also goes to /usr/local/lib (libtest has been created to
> avoid __ctype_b errors because
>    of mismatch between gcc compiler versions of current & the one which
> created libmopac.a
> 7. ON bash shell:
> 	export LDFLAGS="-L/usr/local/lib -Wl,--wrap,__ctype_b"
> 	export CPPFLAGS=-DUSE_MOPAC
> 	export LIBS="-lmopac -ltest"
> 8. in gromacs directory:
> 	./configure --with-qmmm-mopac
> 	make -j4
> 	make install
>
> This completes the compilation w/o any error. I am still to run any qmmm,
> so cann't yet :( comment on the accuracy of process.
>
> --
>
>
>> The trick worked only in gromacs 3.2. You therefore have to use a
>> different restrait, such as distance restraints. Hope this helps.
>>
>> Concerning the cys-cys dimerization, that depends how you do it. I am
>> sorry I can not give a more concise answer to that question. Simplest
>> would be to pull the S atoms together at MM level and perform short QM
>> simulations with restraints on selected frames of the simulatoins.
>> Hope this also helps.
>> Gerrit
>>
>>
>> On 6 Apr 2009, at 13:04, Praveen Agrawal wrote:
>>
>>> Hi Dr.Groenh,
>>> Thankyou for sharing the tutorials on QMMM capabilities of Gromacs,
>>> I am
>>> following the same to be able to study cys-cys dimerization for short
>>> membrane peptide. While following the tutorial I am getting errors
>>> and am
>>> unable to find any other details to overcome it, kindly help:
>>>
>>> 1. ERROR 1 virtual site CD (Res DAT-1) has non-zero mass 12.011
>>>   The 'trick' to create avoid virtual-virtual constraint treats atom
>>> 7 as
>>> virtual which is 'real' atom and has non zero mass. To overcome it I
>>> did->
>>> 2. 32 as combination of 7 &1 and 33 as combination of 16 & 14 but that
>>> leads to the error: Cannot have constraint (32-33) with virtual site
>>> (32)
>>>
>>> I am using Gromacs 4.0.3 and also attaching the corresponding file for
>>> reference. Also, would want to have your opinion on: Is attempting
>>> QM/MM
>>> for cys-cys dimerization for short membrane peptide (25 amino acids
>>> long
>>> each) too far fetch idea?
>>>
>>> Best Regards,
>>> Praveen
>>> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>> http://www.scfbio-iitd.res.in
>>> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~<create_tops.scr>
>>
>> --
>> Gerrit Groenhof
>> MPI biophysical chemistry
>> Goettingen
>> Germany
>> http://wwwuser.gwdg.de/~ggroenh/
>>
>
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> http://www.scfbio-iitd.res.in
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
http://www.scfbio-iitd.res.in
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~