[gmx-users] question about number of atomic contacts beetween residues
feenstra at few.vu.nl
Tue Nov 30 17:14:48 CET 2004
Leonardo Sepulveda wrote:
> Thanks a lot, Anton. I will try to change the g_mdmat code with a
> friend. If it is acomplished i`ll send the code to you.
> Mi "calculations" is the "fraction of contacts (Q) " very used in protein
> folding. I have read that the standard way to do that is calculate the
> Van der Walls radii of each atom plus 0.5 angstroms. If the radii
> (defined in that way) of two atoms overlap, then a contact exist.
> Nevertheless i read in the mindist program (Patricia Soto told me it
> calculate contacts too) that the way to define the radii is giving a
> cutoff (default 6 angtroms) for each atom. Does g_mdmat calculate
> contacts acording to a predefined cutoff or it does acoording to the
> force field van der walls radii of each atom?
It probably uses a cutoff. Caveat here is that there is no such thing as a
'forcefield VdW radius' for an atom. The forcefield defines Lennard-Jones
(LJ) parameters (i.e., sigma & epsilon or C6 & C12). It is not entirely
trivial to convert that to sensible radii.
In the $GMXDATA directory there is a file 'vdwradii.dat' which defines a
set of radii based on atom name (or type, don't know for sure) which you
could use. The programs ngmx and gmxcheck actually use it, you can look
there to see how to do that easiest.
> The paper where i read that metod used CHARMM, so is easy because it
> have all the atoms. How can be done for a united atom FF like
> GROMOS96?? i am confused about it.
Take a molecular viewer (e.g., rasmol) and open any molecule with a CH3
group (all-atom) in it. Display it as 'spacefilling' (or 'CPK'). You'll see
that the hydrogen atoms only extend a little bit beyond the radius of the
carbon. If you also know that hydrogens have a much softer (compared to
carbon) LJ interaction, you can see that it is not a bad approximation to
represent a CH3 group as a carbon atom that is slightly heavier and has a
slightly larger LJ radius.
| | |
| _ _ ___,| K. Anton Feenstra |
| / \ / \'| | | Dept. of Pharmacochem. - Vrije Universiteit Amsterdam |
|( | )| | | De Boelelaan 1083 - 1081 HV Amsterdam - Netherlands |
| \_/ \_/ | | | Tel: +31 20 44 47608 - Fax: +31 20 44 47610 |
| | Feenstra at few.vu.nl - www.few.vu.nl/~feenstra/ |
| | "If You See Me Getting High, Knock Me Down" (RHCP) |
More information about the gromacs.org_gmx-users