#!/usr/bin/perl
# ll_augjob

use strict;
use warnings;
use Cwd;
use Getopt::Long ;
#qw(:config no_ignore_case bundling);

use constant NASV1 => 0;

my $ACCOUNT="none";
my $CLOCKTIME="06:00:00";
my $aug_bin="augustus"; # fixme: param?
my $debug=0;

my( $base_dir, $oldbase, $aug_dir, $extrinsicCfgFile, $species,
  $hintsfile, $genome, $output_file_name, $partitions, $part_dir,
  ) = ("") x 20;

$output_file_name="augustus.gff";
$partitions="partitions.list";
$extrinsicCfgFile="config/augrun.cfg";

# if(NASV1) {
# $species="nasonia_vitripennis";
# $hintsfile="nasv_all.hints";
# $genome="nasv_genome.fa";
# $partitions="nasv_partitions.list";
# $oldbase="/export/home/bio/bio-grid/mb/EVidenceModeler/nasv1/";
# $base_dir="/N/gpfsbr/gilbertd/chrs/aug/nasv1/"; # FIX
# $aug_dir="/N/u/gilbertd/BigRed/bio/augustus";
# }


my $optok= &GetOptions (
  "hintsfile=s" => \$hintsfile,
  "extrinsicCfgFile=s" => \$extrinsicCfgFile,
  "species=s"=>\$species,
  "genome=s"=>\$genome,
  "aug_dir=s"=>\$aug_dir,
  "aug_bin=s" => \$aug_bin,
  "partitions=s"=>\$partitions,
  "output_file_name=s"=>\$output_file_name,
  "ACCOUNT=s"=>\$ACCOUNT,
  "CLOCKTIME=s"=>\$CLOCKTIME,
  "base_dir=s"=>\$base_dir,
  "oldbase=s"=>\$oldbase,
  "debug!"=>\$debug,
  );
  

my $USAGE= <<"USAGE";

usage: ll_augjob --species=nasonia_vitripennis \\
        --aug_dir=/path/to/augustus \\
        --partitions=partitions.list \\
        --genome nasv_genome.fa \\
        --hintsfile=nasv_all.hints --extrinsicCfgFile=aug_dir/config/augrun.cfg \\
        --output augustus.gff \\
optional:
        --ACCOUNT=$ACCOUNT --CLOCKTIME=$CLOCKTIME
USAGE
### need these only if partion on one path, move to another path     
##   --base_dir=/path/to/genome --oldbase=/old/path

# $base_dir = cwd() unless($base_dir);
die "missing --aug_dir=$aug_dir".$USAGE unless (-d $aug_dir);
$partitions= "$base_dir/partitions" unless (-f $partitions);
die "missing --partitions=$partitions".$USAGE unless (-f $partitions);
die $USAGE unless($optok and $species and $genome and $output_file_name);

$aug_dir =~ s,/$,,;
if($hintsfile) {
  $extrinsicCfgFile="${aug_dir}/$extrinsicCfgFile"
    unless($extrinsicCfgFile =~ m,^/,);
  die "missing --extrinsicCfgFile=$extrinsicCfgFile".$USAGE unless (-f $extrinsicCfgFile);
}

my $partn=0;

open (my $fh, $partitions) or die "Error, cannot open $partitions";
while (<$fh>) {
    next unless(/^\w/);
    chomp;
    my ($accession, $acc_dir, $is_partitioned, $partition_dir) = split (/\t/);
    next unless($acc_dir);
    $partn++;
    
    $part_dir = $acc_dir;
    if ($is_partitioned eq "Y") {
        $part_dir = $partition_dir;
    }
    
    $part_dir =~ s,$oldbase,$base_dir, if($oldbase); # die if fail?
    $part_dir =~ s,/$,,;
    
    if (-d $part_dir) {
      submit_augustus($partn, $part_dir);
    } else {
      die "Error, cannot locate data directory $part_dir \n";
    }
 }

# generalize this in package? or instead use ergatis, other workflow tool?
# submit_job_augustus_llsubmit
# submit_job_snap_gridrun

sub submit_augustus {
  my($partn, $part_dir)= @_;
  
  my $jobn="augrun.$partn";
  my $addhints= ($hintsfile) ?
" --hintsfile=${part_dir}/$hintsfile --extrinsicCfgFile=$extrinsicCfgFile "
: "";
 

  my $job = <<"EOF";
#! /bin/bash -l
## $jobn: IBM LoadLeveler job, llsubmit
# @ class = NORMAL
# @ account_no = $ACCOUNT
# @ wall_clock_limit = $CLOCKTIME
# @ error   = $part_dir/$jobn.err
### output  = $part_dir/$jobn.out
# @ notification = always
# @ environment=COPY_ALL;
# @ queue

${aug_dir}/bin/${aug_bin} --gff3=on --uniqueGeneId=true \\
  --species=$species $addhints \\
  --AUGUSTUS_CONFIG_PATH=$aug_dir/config/ \\
  ${part_dir}/$genome  \\
  > ${part_dir}/$output_file_name
EOF
  
  open(J,">$jobn") or die "$jobn"; print J $job; close(J);
if($debug) { # DEBUG off
  print "llsubmit -q $jobn\n";
} else {
  my $res=`llsubmit -q $jobn`; chomp($res);
  print "# llsubmit $jobn : $res\n";
}  
  # unlink $jobn;

}


__END__

parts list
SCAFFOLD20      /export/home/bio/bio-grid/mb/EVidenceModeler/nasv1/SCAFFOLD20   Y       /export/home/bio/bio-grid/mb/EVidenceModeler/nasv1/SCAFFOLD20/SCAFFOLD20_1-2000000
SCAFFOLD20      /export/home/bio/bio-grid/mb/EVidenceModeler/nasv1/SCAFFOLD20   Y       /export/home/bio/bio-grid/mb/EVidenceModeler/nasv1/SCAFFOLD20/SCAFFOLD20_1950001-2035339
SCAFFOLD33      /export/home/bio/bio-grid/mb/EVidenceModeler/nasv1/SCAFFOLD33   N


ll_augjob --species=nasonia_vitripennis \
        --extrinsicCfgFile=$aug_dir/config/augrun.cfg \
        --hintsfile=$base_dir/SCAFFOLD33/nasv_all.hints \
        $base_dir/SCAFFOLD33/nasv_genome.fa \
        > $base_dir/SCAFFOLD33/augustus.gff

>>>>>


ll_augjob \
    --base_dir=$base_dir \
        --species=nasonia_vitripennis \
        --extrinsicCfgFile=$aug_dir/config/augrun.cfg \
        --hintsfile=SCAFFOLD33/nasv_all.hints \
        --genome SCAFFOLD33/nasv_genome.fa \
        --output SCAFFOLD33/augustus.gff