RE: [PyMOL] trouble making movie on linux

[DeLano, W. <wa...@su...>]

At first I thought this was a bug, but after Russel sent me the files, =
it turns out that this reveals an important concept for movie-making =
with PyMOL...

In order to make movies which associate atom identities through a =
trajectory , you must specify unique identifiers for each atom.  Just =
the atomic elements aren't usually good enough since how will PyMOL know =
which carbon in one from is associated with which carbon in the next?  =
Instead, each atom must have a unique combination of =
(segi,chain,resn,resi,name). =20

An easy solution with small molecular systems is to simple load the =
movie into discrete objects, where PyMOL doesn't attempt to associate =
atoms and bonds between states.  This is done with an optional argument =
"discrete=3D1" on the first state which is loaded...

load subnoh00.pdb,mov,discrete=3D1
load subnoh01.pdb,mov
load subnoh02.pdb,mov
load subnoh03.pdb,mov

With discrete objects, if you click on a atom in one state, then PyMOL =
won't know which atom corresponds to what in the next state, and so =
you'll have to manually setup representation within each state if you =
want anything fancy. =20

Also note that with discrete objects, atom and bond information is =
replicated for every state.  In contrast, normal objects share one pool =
of atom, bond, and connectivity information across all states (but not =
every atom needs to appear in each state), and normal objects only store =
unique coordinates for each state.  So, if you're using PyMOL to view =
250 frames of a 50,000 atom MD simulation, discrete objects would be =
memory inefficient.  However, if bonds are made/broken between states, =
then you'll need to use a discrete object, or split the trajectory =
across multiple objects.

Take home lessons:

- For trajectories with abiguous atom names, use discrete objects
- For trajectories where bonds are made/broken, use discrete objects
- For trajectories with static connectivity and unique names, use =
discrete objects.
- For large MD-simulations, use normal objects
- For view docking results, use discrete objects=20

Cheers,
Warren


> I am attempting to make a movie from a series of 50 pdb=20
> files.  When I load=20
> the files individually they work fine, however when I load=20
> the files into a=20
> movie using a series of "load file.pdb,mov" commands only the=20
> first 2 frames=20
> are correct.  In frames 3 and higher the connections between=20
> atoms are drawn=20
> apparently randomly so that the effect is that of a pile of=20
> "pixie sticks"=20
> rather than a molecule.
>=20
> I get this effect on two different machines running redhat=20
> 7.3, one using=20
> pymol 0.78, and another using the "easy install" pymol 0.83=20
> package for=20
> linux.
>=20
> I have though about adding CONNECT records to the pdb files=20
> to force proper=20
> bonding, or perhaps I can use babel to put them into some=20
> other format that=20
> will work.