Thanks again. This indeed works perfectly.
I do have one more question. One of the things my 'process_genome' subroutine has to do is to run a few perl scripts (from JBrowse browser package, if it's of interest).

I did not write those scripts and have no control over them. These scripts require to be run in the "correct" directory (they do stuff in the working directory rather then ask a directory path as an argument). So. my 'process_genome' has to 'chdir' to the genome directory. However, as I learned the current directory is shared by all threads, I was suggested to write the following wrapper:

# forks then changes working directory and executes a command
sub chdir_and_system {
    my ( $dir, @exec_args ) = @_;
    my $pid = fork;
    die "couldn't fork: $!" unless defined $pid;
    unless ($pid) {              # this is the child
        chdir($dir);

       # open STDOUT, ">/dev/null";    # silence subprocess output
       # open STDERR, ">/dev/null"; # silence subprocess stderr
        system(@exec_args) == 0 
            or die "system @exec_args failed";
    }
    wait;
    die "child error $?" if ($?);
}
[download]

then call each of the external scripts using something like this:

sub add_gff_track {
    my ( $browser_project_dir, $gff_file, $track_label, $key, @optiona
+l_args )
        = @_;
    my @args = (
        "bin/flatfile-to-json.pl", "--gff=$gff_file",
        "--tracklabel=$track_label", "--key=$key", @optional_args
    );
    chdir_and_system( $browser_project_dir, @args );

}
[download]

I was originally suggested to use 'exec' in the wrapper, but after some reading I thought that perhaps 'system' should be used. I'm not sure. the thing is I need to know that when this wrapper sub ends, the system command also ended (other command might rely on its output). I'm also not sure if all the wrapper logic is OK. Bottom line, my threads terminate abnormally. Perhaps the JBrowse scripts don't return a good value even if they succeed?

How would you go about this?