removing temp folder

[verweijs]

Hi,

If I run the run_DNSS2.pl script the temp folder is not removed. Looking in the script I see that this line at the sub remove_intermediates is hashed out (line394). Removing the hash didn't solve my problem. Can you help me out?

Cheers and keep up the good work,
Stefan

[verweijs]

The addition of system and $outdir seems to be the key to removing temporary files at line 394:

sub remove_intermediates {
   system "rm -rf $outdir/temp/feat/* $outdir/temp/pfeat/* $outdir/temp/pprob/* $outdir/temp/prob/* $outdir/temp/pssm/*";

[verweijs]

However, since I am trying to run on multiple cores at once to predict multiple proteins (saves a lot of time) this solution won't stand since it removes everything in the general temp directory, which results in the other parallel-operating cores not having this temporary data anymore.

A better solution would be to make a temporary directory per sequence and deleting them after prediction. So in the run_DNSS2.pl script somewhere around line 46 introduce the specific directory, so you would get something like:

my $modeldir = "$GLOBAL_PATH/models/";
my $tempdir = "$outdir/temp/";
my $namedir = "$outdir/temp/$name";
my $pssmdir = "$outdir/temp/$name/pssm/";
my $hmmdir = "$outdir/temp/$name/hmm/";
my $probdir = "$outdir/temp/$name/prob/";
my $featdir = "$outdir/temp/$name/feat/";
my $pfeatdir= "$outdir/temp/$name/pfeat/";
my $pprobdir= "$outdir/temp/$name/pprob/";

`mkdir $outdir` unless (-d $outdir);
`mkdir $tempdir` unless (-d $tempdir);
`mkdir $pssmdir` unless (-d $pssmdir);
`mkdir $hmmdir` unless (-d $hmmdir);
`mkdir $probdir` unless (-d $probdir);
`mkdir $featdir` unless (-d $featdir);
`mkdir $pfeatdir` unless (-d $pfeatdir);
`mkdir $pprobdir` unless (-d $pprobdir);

and then in the remove part you would get something like:

sub remove_intermediates {
   system "rm -rf $outdir/temp/$name/feat/* $outdir/temp/$name/pfeat/* ...
}

However, $name first has to be declared of course..

[huge200890]

This is a good way. Even though the current program supports the input that contains multiple proteins, the prediction will be executed sequentially, If you plan to run program on multiple proteins in parallel, you can also provide different output folders for different proteins, and run them separately. Would this solve the problem?