#!/usr/bin/perl

=item tanyest.pl

 gzcat $sc/dpulex4/dpulex-EST_j06.gff.gz $sc/dpulex4/dpulex-EST_j05.gff.gz \
 | cat $dpe/daph-est-07.info dpulex1_tandy6g.rgn - \
  | env estdetail=1 perl tandyest.pl > dpulex-tandy-estdet.gff
 
=cut

use strict;
my %info;
my ($itab,$gtab,$ttab,@ihead,$debug,$debug2,%rgn);
my $wantgroupdetail= $ENV{estdetail}; # || 0

while(<>){

next if(/^#/ || !/\S/); 
my $in=$_; chomp;  my @v= split"\t";
if(/^LibraryID/ && !@ihead) { $itab=1; %info=(); @ihead= map{ s/\s+$//; s/^\s+//; $_; } @v; }
elsif(/__END__/) { $itab=0; }
elsif(/tandy\tregion/) { $gtab= $itab= 0; $ttab=1; } # tandy region gff
elsif(/tandy\tmatch/) { $gtab= $itab= 0; $ttab=1; } # tandy gene-model gff
elsif(/^scaffold/) { $itab= $ttab= 0; $gtab=1; } ## blastn:EST or Dbxref=JGI:JGI_CBIA392
elsif($itab && @v>5) { 
  my $jid= $v[1]; 
  foreach my $i (0..$#ihead) { $info{$jid}{$ihead[$i]}= $v[$i]; } 
#warn "info($jid)=",join(";", keys(%{$info{$jid}}) ),"\n" if $debug2++ < 30;
#warn "info($jid)=",join(";", values(%{$info{$jid}}) ),"\n" if $debug2 < 30;
}

if($ttab) {
  my($ref,$src,$type,$b,$e,$pr,$or,$ph,$attr)= @v;
  $attr =~ m/cid=(\d+)/; my $c=$1 || 0; 
  my $bn=int(($b+$e)/10000); 
  push @{$rgn{$ref}{$bn}}, [$b,$e,$c]; 
  
} elsif($gtab) {
  my($ref,$src,$type,$b,$e,$pr,$or,$ph,$attr)= @v;
  my($pid,$jid)=("","");
  $attr =~ m/Parent=(WFes\d+)/; $pid= $1;
  ## $attr =~
  $in =~ m/Dbxref=JGI:(\w+)/; $jid= $1; # Dbxref=JGI:JGI_CBIA501.rev

  my $bn= int(($b+$e)/10000); 
   
  if ($type =~ /HSP/ && $src =~ /EST_j/) { ##  ## blastn:EST_j06
    # ($eid = $pid) =~ s/_\w+//; 

#warn "jid=$jid\n" if $debug++ < 20;    
    $jid =~ s/\d+//;  $jid =~ s/JGI_//; 
    $jid =~ s/\.(fwd|rev)//; #? keep
#warn "info($jid)=",join(",", values(%{$info{$jid}}) ),"\n" if $debug<10;
    my $estgroup= jid2env($jid);
    $in =~ s/;/;estgroup=$estgroup;/ if($estgroup);
    
    my $rg= $rgn{$ref}{$bn}; my $ov=0; my $c=0; 
    if($rg){ foreach my $a (@$rg){ 
      $c= $$a[2]; $ov=1 and last if($e >= $$a[0] && $b <= $$a[1]); } }
    $in =~ s/;/;tandy=$c;/ if($c);
    
    $in =~ s/;tloc=[^;\s]+//; $in =~ s/;align=[^;\s]+//; #tloc=253-630;align=378
    print $in;
    }
}

}

sub jid2env {
  my($jid)= @_;
  local $_= $info{$jid}{"Condition,Stage"};   ## Condition,Stage
#warn "$jid => $_\n" if($debug++ < 10);
  return $_ if($wantgroupdetail);
  return "normal" if(/(Unchall|Female|Male)/i);
  return "metal" if(/(metal|cadmium|arsenic|zinc|nickel|copper)/);
  return "biotic" if(/(infect|predat|starv|fed)/);
  return "inorganic" if(/(Hypoxia|UV_exposure|nanopart|Acid|salinity|hormone)/);  
  return undef; #? $jid; ##? undef;
}