VMD-L Mailing List
From: Mayne, Christopher G (cmayne2_at_illinois.edu)
Date: Fri Jul 24 2015 - 09:34:38 CDT
1) The divide and conquer approach is almost always the best way to work with large molecules. I'm not aware of any hard and fast rules for fragmentation. Generally, it is best to avoid breaking conjugated systems or places where dynamics are highly coupled (e.g., cyclic). I also look for places in the molecule where there is good (or complete) coverage by existing parameters or highly analogous parameters as good fragmentation points. Recent improvements to the "Identify Missing Parameters" tool from the BuildPar tab to visually inspect the molecule and where missing parameters are located within the molecule.
2) Yes. ffTK was designed to allow users to fully parameterize or refine all or portions of a molecule, and allows the ability to use initial guesses (e.g., from ParamChem/CGenFF Program). The latter has very recently been improved by allowing the user to provide the input and output from the CGenFF Program to automatically construct the PSF/PDB and initial PAR files; however, it is still part of the alpha builds (http://www.ks.uiuc.edu/Research/vmd/alpha/) and needs testing.
3) The algorithm for the autodetection scheme is pretty rudimentary and I recently described it in more depth here:
The reason that the carbons in the video weren't assigned is because they are all sp3 hybridized, meaning that they have 4 bonded partners and are therefore considered non-accessible. Aromatic carbons, however, only have 3 bonded partners and are accessible from the top and bottom faces. Whether they are donors or acceptors will depend on the partial atomic charge of each carbon, as well as the surrounding charge environment. This is difficult to guess a priori, and therefore, we suggest that users calculate both orientations of water interaction (they're not that expensive) and look at the results.
On Jul 24, 2015, at 8:31 AM, ABEL Stephane 175950 wrote:
> Hello everyone,
> I am a new user of the fftk program and sorry if my question are naive I would like to improve the partial charges and the bonded parameters of a molecule that contains 49 atoms (pdb attached ). My molecule is formed with a polyOH benzene ring attached to a isopreneoid tail (i.e. (CHCH3CH2)2OH). My aim is to model this molecule by using the CGenFF force field. So I have firstly used Paramchem to see if the molecule is fully supported (i.e. w/o changes of the default parameters . Paramchem indicates higher penalties for some bonded parameters and atomic charges (49 and 22, respectively, see the results attached ). In particular some dihedrals have a penalty values around 22 and others around 4. So I am following the video tutorials and apply the fftk different steps to my molecule. Below my questions
> 1) Since the molecule has formed with 2 groups it is possible to use a "divide and conquer" approach (i.e. separate the molecule into two blocks) to speedup the optimization process? If yes how to choose the two molecular blocks (are there some rules we have to follow?)
> 2) In general it is safe to keep some parameters unchanged during the optimization process (those with penalty values ~ 0?
> 3) In the 3rd step (Water Int) I have selected only as donnors and acceptors the Hs and Os, respectively, However if I click on AutoDetct indice button fftk also includes in the list the C atoms as both donnors and acceptors (green sphere). I am confused about this selection since in the video tutorial the C atoms are not include in the charge derivations.
>  https://drive.google.com/file/d/0B3sW6cS-tVa2OFdDM29MM29faGc/view?usp=sharing
>  https://drive.google.com/file/d/0B3sW6cS-tVa2Zk1EeTIzTnNhWG8/view?usp=sharing
> Thanks and sorry again if my questions are naive.
> Stéphane Abel, Ph.D.
> CEA Saclay DSV/IbItec-S/SB2SM & CNRS UMR 9198
> Institut de Biologie Intégrative de la Cellule (I2BC)
> Bat 528, Office 138C
> Gif-sur-Yvette, F-91191 FRANCE
> Phone (portable) : +33 6 49 37 70 60