VMD-L Mailing List

From: Marcos Sotomayor (sotomayo_at_ks.uiuc.edu)
Date: Mon Jan 08 2007 - 09:53:34 CST

Hi Leonardo,

you must use "$sel delete" in order to free the memory used by the
atomselect command. Have a look at the box "Atom selections inside loops"
in the last part of the VMD tutorial
(http://www.ks.uiuc.edu/Training/Tutorials/vmd/tutorial-html/node4.html#SECTION00045000000000000000
)

Saludos,
Marcos

On Mon, 8 Jan 2007, Leonardo Sepulveda Durán wrote:

> Hello. I have 3 systems simutated at different temperatures. For each
> System-Temperature pair I have 20 simulations started from different random
> seeds. I would like to calculate, for each 20-simulation group, the rmsds
> between all trajectory pairs. I wrote a script to do this:
>
> # cycle ovel all simulation temperatures
> foreach temp { 300 400 450 500} {
>
> # Cycle over all systems
> foreach name { paa pac pan} {
>
> # load the 20 trajectories for each temperature and type of system
> for {set i 1} {$i <= 20} {incr i} {
> mol load psf ./$temp/din-$temp-$name-$i.psf dcd
> ./$temp/din-$temp-$name-$i.dcd
> }
>
> # Obtain the index of the top molecule (the last loaded)
> set molnum [molinfo top]
>
> # define a counter for the traj-traj comparisons
> set counter 0
>
> # Main calculation cycle. For each temperature and system, this code
> calcule all the trajectory-
> # trajectory rmsds, and then print the information to independent
> files. The "for" loops decrease their
> # indexes as the provided trajectory number is the last one
> for {set i $molnum} {$i >= [ expr $molnum - 20 + 1 ] } {incr i -1} {
> for {set j [ expr $i - 1] } {$j >= [ expr $molnum - 20 + 1 ] }
> {incr j -1} {
>
> incr counter
>
> set num_frames [molinfo top get numframes]
>
> # make selection with all the atoms
> set all [atomselect $i all]
>
> # Calculate the rmsd between frames k of 2 trajectories
> for {set k 0 } {$k < $num_frames} {incr k } {
>
> # select the structures to be compared
> set sim1 [atomselect $i "protein" frame $k]
> set sim2 [atomselect $j "protein" frame $k]
>
> # do the alignment
> set trans_mat [measure fit $sim1 $sim2]
> $all move $trans_mat
>
> # measure RMSD between frames
> set rms [measure rmsd $sim1 $sim2]
>
> # define rmsd array numbering
> set h [expr $k + 1]
>
> set rmsd($h) $rms
> }
>
> puts "\t$counter $i $j"
>
> # open output file
> set outfile [open ./rmsd/$name-$temp-$counter.dat w]
> for {set m 1} {$m <= $num_frames} {incr m} {
> puts $outfile [format "%5d%7.3f" $m $rmsd($m)]
> }
> close $outfile
> }
> }
>
> unset sim1
> unset sim2
> unset all
> unset rmsd
> unset rms
> unset h
>
> # delete all the loaded trajectories
> mol delete all
>
> }
> }
>
> I put the script above in a file called rmsdv1.0.tcl, and ran the script
> with the command "vmd -dispdev text -e rmsdv1.0.tcl".
>
> It works without any trouble when I use 5 simulations instead of 20. In the
> later case, the calculation dies out without any explanation besides writing
> "killed" in the console. During the calculation, each iteration becomes
> slower, so I think that the problem is asociated to memory usage. I dont
> know how the memory is used during the computation, but is reasonable that
> all the iterations where calculated at the same rate. It seems that after a
> pairwise comparison the compared simulations remain in memory, I tried to
> solve the problem using the "unset" statements at the end of the script, but
> it seems to have no effect at all. Maybe the problem is another, any help
> from a tcl programmer could help me.
>
> Thanks in advance.
>
> Leonardo Sepulveda
> Universidad de Chile
>