AMBER Archive (2004)

Subject: Re: AMBER: problems with dihedrals - extended information

From: Petr Kulhanek (kulhanek_at_chemi.muni.cz)
Date: Sun Oct 17 2004 - 10:39:31 CDT


> I'm not sure what you mean by "number of dihedrals". Can you
> be specific as to what value in the prmtop you are using for this?
> Is it dihedrals involving hydrogen, not involving hydrogen, or number
> of dihedral types, etc. one complication that you may already understand
> is that dihedrals are given pointers to dihedral types, which actually
> have the parameters. it's not surprising that the number of types doesn't
> sum up, since some may be shared. the actual numbers themselves
> (such as NPHIH, MPHIA) should sum up fine.
> if your program can extract part of the prmtop to make a new prmtop,
> can you compare the dihedrals listed in each prmtop and find out
> which ones differ?

For ligand parm (parm A - ) extracted from complex one, the info about
dihedrals is as follows:
  Number of dihedrals : 161
  Number of dihedrals with hydrogens : 95
  Number of dihedrals without hydrogens: 66
  Number of dihedral types : 22

For ligand parm (parm B) build directly in tleap, the info about dihedrals
is as follows:
  Number of dihedrals : 168
  Number of dihedrals with hydrogens : 101
  Number of dihedrals without hydrogens: 67
  Number of dihedral types : 22

I went deeply to the problem and I found following:
a) in both parm files, the dihedral types are exactly same but in different
order.
b) parm B contains the same number of dihedrals containing hydrogens but 6
of them have two terms instead of single term as in parm A.
c) parm B contains the same number of dihedrals not containg hydrogens but
one of them has three terms instead of two terms as in parm A.

The list of problematic dihedrals is following:

containing hydrogens
atom indexes (regular counting from zero) - atom names - atom types
35 34 33 39 - H07 C02 C06 O15 - H1-CT-CT-OH
25 24 26 29 - H09 C04 C11 O19 - H1-CT-CT-OH
23 22 34 36 - H03 C01 C02 O08 - H1-CT-CT-OH
12 11 13 16 - H36 C30 C37 O42 - H1-CT-CT-OH
04 02 01 06 - O28 C21 C29 H33 - OH-CT-CT-H1
03 02 01 07 - H27 C21 C29 O34 - H1-CT-CT-OH

not containing hydrogens
atom indexes (regular counting from zero) - atom names - atom types
04 02 01 07 - O28 C21 C29 O34 - OH-CT-CT-OH

in parm A, the dihedrals of type H1-CT-CT-OH have assigned following
parameters
PK PN PHASE
0.25000 1.0 0.0

in parm B, the dihedrals of type H1-CT-CT-OH have assigned following
parameters
PK PN PHASE
0.25000 1.0 0.0
0.15556 3.0 0.0

in parm A, the dihedral of type OH-CT-CT-OH has assigned following
parameters
PK PN PHASE
-0.67000 1.0 0.0
0.500000 2.0 0.0

in parm B, the dihedral of type OH-CT-CT-OH has assigned following
parameters
PK PN PHASE
-0.67 1.0 0.0
0.5 2.0 0.0
0.15556 3.0 0.0

Above mentioned dihedral types match following parameters from both
parm99.dat and glycam_2000a.dat files.
(parm99.dat)
X -CT-CT-X 9 1.40 0.0 3.
H1-CT-CT-OH 1 0.25 0.0 1.
OH-CT-CT-OH 1 0.144 0.0 -3.
OH-CT-CT-OH 1 1.175 0.0 2.

(glycam_2000a.dat)
X -CT-CT-X 9 1.40 0.00 3.
OH-CT-CT-OH 1 -0.67 0.00 -1.
              1 0.50 0.00 2.

>From these information follows that in the case of parm B tleap uses both
specific and generic dihedral parameters
while in the case of parm A tleap uses only specific parameters.
How is this possible ? Is this error in tleap or did I do some mistake ?

At the end I must mention that in the archive with supplemetary files
(http://www.ncbr.chemi.muni.cz/~kulhanek/downloads/top_tests.tar.gz), the
parm A (ext_ligand.top)
was not correctly extracted from complex topology (parm file did not provide
correct values of 1-4 energy terms in subsequent calculation). This is now
fixed. When I was writing the program I strictly followed the description of
parm file (amber manual) however sander 8 does not follow it as strictly as
me (I will post this as separate message.)

Thanks in advance for any answer,
        Petr

##########################################
 RNDr. Petr Kulhanek
 National Centre for Biomolecular Research
 Faculty of Science, Masaryk University Brno
 Kotlarska 2, 611 37 Brno, Czech Republic
 e-mail: kulhanek_at_chemi.muni.cz
##########################################

> ===================================================================
> Carlos L. Simmerling, Ph.D.
> Associate Professor Phone: (631) 632-1336
> Center for Structural Biology Fax: (631) 632-1555
> Stony Brook University Web: http://comp.chem.sunysb.edu/carlos
> Stony Brook, NY 11794-5115 E-mail: carlos.simmerling_at_stonybrook.edu
> ===================================================================
>
>
>
>
> Petr Kulhanek wrote:
>
> >Dear Amber Community,
> >
> >I did MM/PBSA analysis of protein + ligand complex from single trajectory
> >and I found that the dihedral energy contribution to interaction energy
was
> >not zero. So I tried to locate the stage from the mistake or mistakes
could
> >come and I found out that the ligand topology contains different number
of
> >dihedrals than the same part in complex topology.
> >
> >Thefore I prepared toplogies once again with the exactly same order of
prep
> >and parameter files in tleap using following script:
>
>---------------------------------------------------------------------------
-
> >--------
> >source leaprc.ff99
> >loadamberparams glycam_2000a.dat
> >loadamberparams mangan.dat
> >loadamberparams pyr.dat
> >loadamberprep la.prep
> >loadamberprep mangan.prep
> >loadamberprep pyr.prep
> >loadamberprep uracil.prep
> >loadamberprep deoxylac.prep
> >
> >COMPLEX=loadpdb complex.pdb
> >desc COMPLEX.285
> >desc COMPLEX.286
> >bond COMPLEX.285.1 COMPLEX.286.1
> >saveamberparm COMPLEX complex.top complex.crd
> >
> >PROTEIN=loadpdb protein.pdb
> >desc PROTEIN.285
> >desc PROTEIN.286
> >bond PROTEIN.285.1 PROTEIN.286.1
> >saveamberparm PROTEIN protein.top protein.crd
> >
> >LIGAND=loadpdb ligand.pdb
> >saveamberparm LIGAND ligand.top ligand.crd
> >
> >quit
>
>---------------------------------------------------------------------------
-
> >-------
> >
> >The results are as follows:
> >
> > Number of dihedrals Dihedral energy
> >protein 16957 2308.3250
> >ligand 168 7.8991
> >sum 17125 2316.2241
> >---------------------------------------------------------
> >complex 17118 2316.1308
> >======================================
> >difference 7 0.0933
> >
> >The results confirm that the ligand contains different number of
dihedrals
> >which results to non-zero dihedral energy contribution to interaction
> >energy.
> >
> >To bring more light in my problem I wrote program which is capable to
> >extract any part from topology as new one. If I use ligand topology
> >(ext_ligand) which is extracted by this program from complex topology the
> >results are now correct.
> >
> > Number of dihedrals Dihedral energy
> >protein 16957 2308.3250
> >ext_ligand 161 7.8058
> >sum 17118 2316.1308
> >---------------------------------------------------------
> >complex 17118 2316.1308
> >======================================
> >diff 0 0.0000
> >
> >Unsolved question for me is 'how is it possible that the the number of
> >dihedrals of some individual molecule depends on the rest of system?'.
> >
> >Some additional notes:
> >tleap from AMBER 8 distribution was used.
> >The other aditive energy contributions to interaction energy (bond and
> >angle) are zero in all cases.
> >Ligand is deoxylactose and protein contains galactosyl-UDP.
> >All files related to this problem can be downloaded from:
> >http://www.ncbr.chemi.muni.cz/~kulhanek/downloads/top_tests.tar.gz
> >
> >Thanks in advance for any answer,
> >
> > Petr Kulhanek
> >
> >##########################################
> > RNDr. Petr Kulhanek
> > National Centre for Biomolecular Research
> > Faculty of Science, Masaryk University Brno
> > Kotlarska 2, 611 37 Brno, Czech Republic
> > e-mail: kulhanek_at_chemi.muni.cz
> >##########################################
> >
> >-----------------------------------------------------------------------
> >The AMBER Mail Reflector
> >To post, send mail to amber_at_scripps.edu
> >To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
> >
> >
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu