AMBER Archive (2009)

Subject: Re: AMBER: a question about ff03ua

From: FyD (fyd_at_q4md-forcefieldtools.org)
Date: Fri Jan 02 2009 - 14:20:37 CST

Dear Oguz Gurbulak,

> For example, for n-dodecane, I tried to generate a mol2 file by hand. I
> attached the mol2 file to the mail. I arraged the charges for C-H and
> C-H-H interactions and then I erased H atoms. All the charges in the
> file is zero.

I looked at your mol2 file: I do not understand the charge value
equivalencing procedure for chemically equivalent atoms, the atom
connectivities are strange & the last section of your mol2 file is not
present...

I ran 2 quick jobs using R.E.D.Server/R.E.D.-IV (based on
HF/6-31G*//HF/6-31G*) for n-dodecane,
http://en.wikipedia.org/wiki/Dodecan; using a single conformation
('extended') among the 355 theoretically possible ones (dixit
wikipedia) & 2 molecular orientations...

R.E.D.-IV has generated a united atom (UA) FF library versus the
non-united atom FF library version. Data generated using R.E.D.-IV in
two different runs. Then, I loaded the UA FF library in LEaP:

ln -s Mol_m1-o1.mol2 Doc.mol2

[fydred_at_node0 UA-FF]$ tleap
-I: Adding /usr/local/amber10-mkl/dat/leap/prep to search path.
-I: Adding /usr/local/amber10-mkl/dat/leap/lib to search path.
-I: Adding /usr/local/amber10-mkl/dat/leap/parm to search path.
-I: Adding /usr/local/amber10-mkl/dat/leap/cmd to search path.
Welcome to LEaP!
(no leaprc in search path)
> DOC = loadmol2 Doc.mol2
Loading Mol2 file: ./Doc.mol2
Reading MOLECULE named DOC
> charge DOC
Total unperturbed charge: 0.000000
Total perturbed charge: 0.000000

The 2 R.E.D.-IV runs are @ http://q4md-forcefieldtools.org/FyD/Dodecan.tgz

R.E.D.-IV log files:
http://q4md-forcefieldtools.org/FyD/Dodecan/RED-v4-Dec08-Not-UA-FF.log
http://q4md-forcefieldtools.org/FyD/Dodecan/RED-v4-Dec08-UA-FF.log

Directory for charge derivation & FF library building:
http://q4md-forcefieldtools.org/FyD/Dodecan/Not-UA-FF/
http://q4md-forcefieldtools.org/FyD/Dodecan/UA-FF/

Tripos mol2 files:
http://q4md-forcefieldtools.org/FyD/Dodecan/Not-UA-FF/Mol_m1-o1.mol2
http://q4md-forcefieldtools.org/FyD/Dodecan/UA-FF/Mol_m1-o1.mol2

Finally, I would strongly suggest you to read:
http://www3.interscience.wiley.com/journal/109583172/abstract?CRETRY=1&SRETRY=0
Failure of net atomic charge models to represent the van der Waals
envelope electric potential of n-alkanes, Donald E. Williams
D.E. Williams J. Comput. Chem. 1994, 15, 719-732.

Indeed, the RRMS of the fit is very bad:
tail -n 1 output2_m1
   ESP relative RMS (SQRT(chipot/ssvpot)) 0.93066

This problem has been reported several times. This is also described
in R.E.DD.B. for cyclohexane http://en.wikipedia.org/wiki/Cyclohexane.
See http://q4md-forcefieldtools.org/REDDB/uploadfile/W-43/

I hope this helps,
regards, Francois

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
      to majordomo_at_scripps.edu