#!/usr/bin/perl -w if(scalar(@ARGV) < 2) { print "runAmemeMuscle.pl - Run the improbizer lots of times for a series of files of interest. Assumes that they have the 'up' vs. 'down' naming convention for the 3' and 5' intronic sequences respectively. Uses the appropriate control as a background distribution. usage: runAmeme.pl outputSummary.html file1Inc.fa file2Inc.fa ... "; exit(1); } $ameme = "/cluster/home/sugnet/bin/i386/ameme"; $upBackground = "fa/mm2.control.gold.up.fa"; $downBackground = "fa/mm2.control.gold.down.fa"; $file = shift(@ARGV); open(HTML, ">$file"); print HTML "\n"; $htmlDir = 'html/'; $tmpDir = 'tmp/'; $numControls = 100; $controlDir = 'controls/'; print HTML "\n"; while($goodFile = shift(@ARGV)) { print HTML ""; chomp($goodFile); $ave = 0; $min = 1000000000; $max = -1000000000; print STDERR "Doing $goodFile\n"; $prefix=`basename $goodFile`; chomp($prefix); if($goodFile =~ /(.*)up(.*)/) { $background = $upBackground; } elsif($goodFile =~ /(.*)down(.*)/) { $background = $downBackground; } else { die "Can't parse 'up' or 'down' out of $goodFile\n"; } $command = "$ameme good=$goodFile numMotifs=4 background=m2 maxOcc=1 motifOutput=$htmlDir$prefix.motif html=$htmlDir$prefix.html gif=$htmlDir$prefix.gif constrainer=6 bad=$background startScanLimit=1000 2>&1 /dev/null"; `$command`; @p = split /\s/, `head -n 1 $htmlDir$prefix.motif`; chomp($p[5]); print HTML ""; print STDERR "Doing controls"; for($i=0; $i<$numControls; $i++) { print STDERR "."; `rm -f $controlDir$prefix.controls`; $command = "$ameme good=$goodFile numMotifs=4 background=m2 maxOcc=1 motifOutput=$tmpDir$prefix.tmp.motif html=$tmpDir$prefix.tmp.html gif=$tmpDir$prefix.tmp.gif constrainer=6 bad=$background controlRun=on 2>&1 /dev/null"; `$command`; `head -n 1 $tmpDir$prefix.tmp.motif >> $controlDir$prefix.controls`; @p = split /\s/, `head -n 1 $tmpDir$prefix.tmp.motif`; if($p[0] >= $max) { $max = $p[0]; } if($p[0] <= $min) { $min = $p[0]; } $ave += $p[0]; } print STDERR "\n"; $ave = $ave / $numControls; print HTML "\n"; } print HTML "
FileScoremotifControls
$goodFile$p[0]$p[5]$ave ($min-$max N=$numControls)
\n";