pymol_View the correlations in the structure_does_NOT_work

[karolk]

Hi,

I am having some weird problems with pymol. When I invoke it and try to plot correlations on my structure, it simply stucks. The process is running for hours and hours, but the pymol window does not appear. The structure is really large, so I understand that it might take some time, but I print correlations based on the first 30 modes only, so it should work.

The same command (pymol(cm, PDB, type="launch")) works perfectly with tutorial files (1sx4).

So, it is not an issue of pymol, it is not a memory issue (I think), since I have 3-4GB left. I have no idea. Any help will be greatly appreciated.

cheers

[Lars Skjærven]

Try increasing the omit argument from 0.2 to 0.5 (or even higher).

The number of modes included in the calculation doesn't matter. It's the number of residues that determines the dimensions of the correlation matrix.

[karolk]

ok. I will try. thanks

[karolk]

Hi,

Sorry for delayed reply.

No. It is not resolved.

I was trying different omit values.

Pymol finally launched with omit = 0.8, but it is not helpful at all.

With value 0.5 or 0.3 it does not launch even when waiting more than 4-5 hours.

Is there any other way to plot correlations on structure? For example, using VMD?

thanks

[Lars Skjærven]

whats the dimensions of your correlation matrix? any way you could provide some code so that we can reproduce the problem?

[karolk]

4684 per 4684

Currently I don't have access to my desktop. I will send you a code on Monday.

[Lars Skjærven]

VMD is certainly an option that might handle it better. Another option is to filter more of you matrix elements. e.g. you can filter on contact map, and/or cij value. See below an example:

pdb <- read.pdb("1hel")
pdb <- trim(pdb, "protein")
modes <- nma(pdb)
cij <- dccm(modes)
cm <- cmap(pdb)
cij2 <- filter.dccm(cij, cutoff.cij=0.5, cmap=cm)
pymol(cij2, pdb)

see help(cmap) for an overview on what this function returns. e.g. by default all adjacent residues will be filtered out- which might be good for you purpose as you can assume that these will be strongly correlated and there is no point in visualizing all those pairs.

[karolk]

Thanks. I will try and I will let you know.

[karolk]

It worked. You can close this ticket. thanks