VMD-L Mailing List

From: John Stone (johns_at_ks.uiuc.edu)
Date: Tue Aug 05 2008 - 00:36:59 CDT

David,
  You're deleting them too late. The delete should be placed
at the end of the same loop in which you created the selection,
not in the outer scope that contains the loop.
Loops with atom selections should always look something like:

for ... {
  set mysel [atomselect ...]
  do stuff...
  $mysel delete
}

If you wait to delete until after the end of that loop,
it's too late, and you've already "leaked" the memory as
$mysel will get overwritten with a handle to a completely
different selection the next time through the loop.

You can see if you're leaking selections by doing "atomselect list"
after your code runs. Try it on a small test case after you fix
your code to verify you've cleaned things up.

Cheers,
  John Stone
  vmd_at_ks.uiuc.edu

On Tue, Aug 05, 2008 at 03:24:10PM +1000, David Poger wrote:
> Thanks for the quick reply. Unfortunately, I still have a memory leak...
> I added the commands "$jsel delete" and "unset jsel" after the "j" loop
> and "$isel delete" and "unset isel" after the "i" loop though. Should I
> delete the frames once the coordinates of each one has been read? If so,
> how?
>
> Thanks for any help
>
> David
>
>
> for {set i $Dom1_1 } { $i<=$Dom1_n } {incr i} {
> # Number of atoms in the residues of Dom1 which is selected
> set isel [atomselect top "resid $i"]
>
> for {set j $Dom2_1 } { $j<=$Dom2_n } {incr j} {
> set distm 1000
>
> foreach coordi [$isel get {x y z}] {
> # Number of atoms in the residues of Dom2 which is selected
> set jsel [atomselect top "resid $j"]
>
> foreach coordj [$jsel get {x y z}] {
> set dist [vecdist $coordi $coordj]
> if {$dist<$distm} {set distm $dist}
> }
> }
> # The array is in angstroem!!
> set sumdistij($i,$j) [expr $sumdistij($i,$j)+$distm]
> set sumdistij2($i,$j) [expr $sumdistij2($i,$j)+$distm*$distm]
>
> #Release memory!
> $jsel delete
> unset jsel
> }
> $isel delete
> unset isel
> }
>
>
>
>
>
> John Stone wrote:
> >Hi,
> > You must delete the atom selections that you're creating within
> >the loops in your script. Specifically, you must add
> >"$isel delete" and "$jsel delete" (for example) to your nested
> >loops. This is the cause of your memory consumption problem.
> >This is a frequently asked question, it comes up almost once every
> >week or two.
> >
> >Cheers,
> > John Stone
> > vmd_at_ks.uiuc.edu
> >
> >On Tue, Aug 05, 2008 at 11:31:10AM +1000, David Poger wrote:
> >
> >>Hello
> >>
> >>I have written a Tcl script that basically reads a trajectory and for
> >>each frame, finds the shortest distance between every single residue of
> >>a first domain and every single residue of another domain in a protein.
> >>The final matrix with all the minimal distances is written in an output
> >>file. The script works fine... for 2 frames. In fact, after a few
> >>minutes (~10-20 frames), I have to kill it as it fills up the memory of
> >>the PC (32GB of RAM!!!). Is there any way to optimize this script
> >>(release memory for example?). I know I have several nested loops but I
> >>don't how to do differentlty and I have absolutely no knowledge in code
> >>optimization!
> >>
> >>
> >>Thanks a lot!
> >>David
> >>
> >>
> >># In order to run it start vmd without graphical interface by typing
> >># $> vmd -dispdev text -e script.tcl
> >>
> >>#======================================================#
> >># TO BE EDITED #
> >>
> >>set grofile protein.gro
> >>set trajfile traj.xtc
> >>set outputfile traj_contact.dat
> >>
> >># #
> >>#======================================================#
> >>
> >>
> >>set datafile [open $outputfile w]
> >>puts $datafile [format "%1s%7s%8s%9s%9s" "#" "Res1" "Res2" "<d>" "rmsd"]
> >>
> >>
> >>mol new $grofile
> >>
> >># Residues in domains 1D and 2D
> >>set Dom1_1 20
> >>set Dom1_n 155
> >>set Dom2_1 126
> >>set Dom2_n 244
> >>
> >>for {set i $Dom1_1 } { $i<=$Dom1_n } {incr i} {
> >> for {set j $Dom2_1 } { $j<=$Dom2_n } {incr j} {
> >> set sumdistij($i,$j) 0
> >> set sumdistij2($i,$j) 0
> >> }
> >>}
> >>
> >>#animate style once
> >>mol addfile $trajfile waitfor all
> >>set nframes [molinfo top get numframes]
> >>puts "nframes=$nframes"
> >>
> >># The 1st frame is the $grofile used for the topology so I don't use it
> >># That's why k starts from 1 and not 0
> >>for {set k 1} {$k<$nframes} {incr k} {
> >> animate goto $k
> >> display update
> >> puts "frame #$k"
> >>
> >> for {set i $Dom1_1 } { $i<=$Dom1_n } {incr i} {
> >> # Number of atoms in the residues of Dom1 which is selected
> >> set isel [atomselect top "resid $i"]
> >>
> >> for {set j $Dom2_1 } { $j<=$Dom2_n } {incr j} {
> >> set distm 1000
> >>
> >> foreach coordi [$isel get {x y z}] {
> >> # Number of atoms in the residues of Dom2 which is selected
> >> set jsel [atomselect top "resid $j"]
> >>
> >> foreach coordj [$jsel get {x y z}] {
> >> set dist [vecdist $coordi $coordj]
> >> if {$dist<$distm} {set distm $dist}
> >> }
> >> }
> >> # The array is in angstroem!!
> >> set sumdistij($i,$j) [expr $sumdistij($i,$j)+$distm]
> >> set sumdistij2($i,$j) [expr $sumdistij2($i,$j)+$distm*$distm]
> >>
> >> }
> >> }
> >>
> >>}
> >>
> >>
> >>for {set i $Dom1_1 } { $i<=$Dom1_n } {incr i} {
> >> for {set j $Dom2_1 } { $j<=$Dom2_n } {incr j} {
> >> set mean [expr $sumdistij($i,$j)/($nframes-1)]
> >> set mean2 [expr $sumdistij2($i,$j)/($nframes-1)]
> >> set msd [expr $mean2-$mean*$mean]
> >> set rmsd [expr sqrt($msd)]
> >>
> >> puts $datafile [format "%8d%8d%9.3f%9.3f" $i $j [expr $mean/10]
> >>[expr $rmsd/10]]
> >> }
> >>}
> >>
> >>close $datafile
> >>
> >>puts "Finished"
> >>
> >>exit
> >>
> >>--
> >>
> >>_________________________________________________
> >>David POGER, Ph.D.
> >>Postdoctoral Research Fellow
> >>Molecular Dynamics Group
> >>School of Molecular and Microbial Sciences (SMMS)
> >>Chemistry Building (#68)
> >>The University of Queensland
> >>Brisbane, QLD 4067
> >>Australia
> >>
> >>Email: d.poger_at_uq.edu.au
> >>Tel: +61 7 3365 7562
> >>Fax: +61 7 3365 3872
> >>
> >>
> >>** Notice: Unless stated otherwise, this e-mail represents only the views
> >>of the Sender and not the views of The University of Queensland.
> >>
> >
> >
>
> --
>
> _________________________________________________
> David POGER, Ph.D.
> Postdoctoral Research Fellow
> Molecular Dynamics Group
> School of Molecular and Microbial Sciences (SMMS)
> Chemistry Building (#68)
> The University of Queensland
> Brisbane, QLD 4067
> Australia
>
> Email: d.poger_at_uq.edu.au
> Tel: +61 7 3365 7562
> Fax: +61 7 3365 3872
>
>
> ** Notice: Unless stated otherwise, this e-mail represents only the views
> of the Sender and not the views of The University of Queensland. 

-- 
NIH Resource for Macromolecular Modeling and Bioinformatics
Beckman Institute for Advanced Science and Technology
University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801
Email: johns_at_ks.uiuc.edu                 Phone: 217-244-3349
  WWW: http://www.ks.uiuc.edu/~johns/      Fax: 217-244-6078