#!/usr/bin/perl 
# Script to automate iterative runs of CRITICA

use Critica;

$scriptname=getScriptName();
@files=realArguments(@ARGV);

(@files==3) || usage();

$fraction=0.8;
$iterations=3;
$addlong=0;
$nosd=0;
$noprom=1;
$startiteration=0;
$dicodonscores="";
$initscores="";
$options="";
$fixedoptions="";
$gencode="";

process_options(flagArguments(@ARGV));

$tripletsfile=pop(@files);
$seqfile=pop(@files);
$outputprefix=pop(@files);

if(!-e($seqfile)) {
  die("$scriptname: No such file as $seqfile!\n");
}
if(!-e($tripletsfile)) {
  die("$scriptname: No such file as $tripletsfile!\n");
}

$totalinits=$outputprefix.".init";


for($i=$startiteration;$i<=$iterations;$i++) {
  $options.=$fixedoptions;
  $scratch=$outputprefix.".tmp";
  $criticaoutput=$outputprefix.$i;
  $crit=$criticaoutput.".crit";
  $mesg=$criticaoutput.".mesg";
  $cdsfile=$criticaoutput.".cds";
  $rejectfile=$criticaoutput.".reject";
  $sdscores=$criticaoutput.".sdscores";
  $promscores=$criticaoutput.".promscores";
  $dicodonscores=$criticaoutput.".dicod";
  $initscores=$criticaoutput.".init";
  $err=system("critica $seqfile $tripletsfile $gencode $options > $crit 2> $mesg");
  if ($err) {
    printf("critica died with error message %d\n",$err/256);
    die("Aborting\n");
  }
  bestInits($crit,$scratch);
  if (($i==0) && ($addlong>0)) {
    $err=system("addlongorfs -orf-aa-length=$addlong $gencode $scratch $seqfile >> $scratch");
    if ($err) {
      printf("addlongorfs died with error message %d\n",$err/256);
      die("Aborting\n");
    }
  }
  $err=system("removeoverlaps $seqfile $scratch 20 > $cdsfile 2>$rejectfile");
  if ($err) {
    printf("removeoverlaps died with error message %d\n",$err/256);
    die("Aborting\n");
  }
  cdsSort($cdsfile);
  if ($i<$iterations) {
    if (!$nosd) {
      computeSDScores($crit,$sdscores);
    }
    if (!$noprom) {
      computePromScores($crit,$promscores);
    }
    computeInitScores($cdsfile,$seqfile,$initscores);
    if ($i==0) {
      $diflag="-fraction-coding=$fraction";
    }
    else {
      $diflag="";
    }
    $err=system("dicodontable $diflag $cdsfile $seqfile > $dicodonscores 2>> $mesg");
    if ($err) {
      printf("dicodontable died with error message %d\n",$err/256);
      die("Aborting\n");
    }
    $options="-dicodon-scores=$dicodonscores -init-scores=$initscores";
    if (!$nosd) {
      $options.=" -sd-scores=$sdscores";
    }
    if (!$noprom) {
      $options.=" -prom-scores=$promscores";
    }

  }
}
$criticaoutput=$outputprefix.$i;
$cdsfile=$criticaoutput.".cds";
unlink $scratch;

sub usage {
  printf("Reference: Badger, Jonathan H. and Gary J. Olsen. CRITICA:\n");
  printf("Coding Region Identification Tool Invoking Comparative Analysis.\n");
  printf("Molecular Biology and Evolution 16: 512-524 (1999)\n\n");

  printf("$scriptname: [options] output-name seq-file triplets-scores\n");
  printf("Valid $scriptname options:\n");
  printf("\t-iterations=number (default: 3)\n");
  printf("\t-scoring-matrix=file\n");
  printf("\t-initial-dicod=file\n");
  printf("\t-initial-init=file\n");
  printf("\t-initial-sd=file\n");
  printf("\t-initial-prom=file\n");
  printf("\t-no-sdscores\n");
  printf("\t-prom-find\n");
  printf("\t-threshold=value\n");
  printf("\t-alpha=value\n");
  printf("\t-fraction-coding=value\n");
  printf("\t-add-longorfs=length\n");
  printf("\t-genetic-code=number\n");
  printf("\t-strict-threshold\n"); 
  printf("\t-frameshift-threshold=value\n"); 
  die("\t-quick-stats\n"); 
}

sub process_options {
  foreach $arg (@_) {
    if (index($arg,"-iterations=")>-1) {
      ($dummy,$iterations)=split("=",$arg);
    }
    elsif (index($arg,"-scoring-matrix=")>-1) {
      $fixedoptions.=" $arg";
    }
    elsif (index($arg,"-initial-dicod=")>-1) {
      ($dummy,$dicodonscores)=split("=",$arg);
      $startiteration=1;
      $options.="-dicodon-scores=$dicodonscores";
    }
    elsif (index($arg,"-initial-init=")>-1) {
      ($dummy,$initscores)=split("=",$arg);
      $startiteration=1;
      $options.=" -init-scores=$initscores";
    }
    elsif (index($arg,"-initial-sd=")>-1) {
      ($dummy,$sdscores)=split("=",$arg);
      $startiteration=1;
      $options.=" -sd-scores=$sdscores";
    }
    elsif (index($arg,"-initial-prom=")>-1) {
      ($dummy,$promscores)=split("=",$arg);
      $startiteration=1;
      $options.=" -prom-scores=$promscores";
    }
    elsif (index($arg,"-no-sdscores")>-1) {
      $nosd=1;
      $fixedoptions.=" $arg";
    }
    elsif (index($arg,"-prom-find")>-1) {
      $noprom=0;
      $fixedoptions.=" $arg";
    }
    elsif (index($arg,"-threshold=")>-1) {
      $fixedoptions.=" $arg";
    }
    elsif (index($arg,"-alpha=")>-1) {
      $fixedoptions.=" $arg";
    }
    elsif (index($arg,"-quick-stats")>-1) {
      $fixedoptions.=" $arg";
    }
    elsif (index($arg,"-frameshift-threshold=")>-1) {
      $fixedoptions.=" $arg";
    }
    elsif (index($arg,"-genetic-code=")>-1) {
      $gencode=$arg;
    }
    elsif (index($arg,"-strict-threshold")>-1) {
      $fixedoptions.=" -strict-threshold";
    }
    elsif (index($arg,"-add-longorfs=")>-1) {
      ($dummy,$addlong)=split("=",$arg);
    }
    elsif (index($arg,"-fraction-coding=")>-1) {
      ($dummy,$fraction)=split("=",$arg);
    }
    else {
      die("$scriptname: option $arg not valid!\n");
    }
  }
}