| 1 |
#!/usr/bin/perl |
| 2 |
use strict; |
| 3 |
use Getopt::Std; |
| 4 |
use FindBin;use lib $FindBin::Bin; |
| 5 |
#raw translation of btab into gff3 format |
| 6 |
|
| 7 |
my $usage = q/Usage: |
| 8 |
btab2gff [-R] <btab_input> |
| 9 |
|
| 10 |
-R will swap roles between target and reference |
| 11 |
/; |
| 12 |
umask 0002; |
| 13 |
getopts('R') || die($usage."\n"); |
| 14 |
my $rswap=$Getopt::Std::opt_R; |
| 15 |
|
| 16 |
my %qh; |
| 17 |
|
| 18 |
while (<>) { |
| 19 |
chomp; |
| 20 |
my @t=split("\t"); |
| 21 |
my @hsp=split('~',$t[21]); |
| 22 |
my $qcov=$rswap ? '' : $t[22]; #not there unless we ran btabflt.pl |
| 23 |
my @ex; |
| 24 |
my $strand; |
| 25 |
my ($gseqid, $qid, $qlen, $qs, $qe) = $rswap ? ($t[0], $t[5], $t[18], $t[8], $t[9]) : |
| 26 |
($t[5], $t[0], $t[2], $t[6], $t[7]); |
| 27 |
my $qn=++$qh{$qid}; |
| 28 |
my $qname=$qid; |
| 29 |
$qid.=".bt$qn"; |
| 30 |
if ($t[17] eq 'Minus') { |
| 31 |
$strand='-'; |
| 32 |
if ($rswap) { |
| 33 |
@ex= map { my @r=(m/([\d\.]+)/g);[$r[0],$r[1],$r[3],$r[2],$r[4]] } @hsp; |
| 34 |
} |
| 35 |
else { |
| 36 |
@ex= map { my @r=(m/([\d\.]+)/g);[$r[3],$r[2],$r[0],$r[1],$r[4]] } @hsp; |
| 37 |
} |
| 38 |
} |
| 39 |
else { |
| 40 |
$strand='+'; |
| 41 |
if ($rswap) { |
| 42 |
@ex= map { my @r=(m/([\d\.]+)/g);[$r[0],$r[1],$r[2],$r[3],$r[4]] } @hsp; |
| 43 |
} |
| 44 |
else { |
| 45 |
@ex= map { my @r=(m/([\d\.]+)/g);[$r[2],$r[3],$r[0],$r[1],$r[4]] } @hsp; |
| 46 |
} |
| 47 |
} |
| 48 |
@ex= sort { $main::a->[0] <=> $main::b->[0] } @ex; |
| 49 |
($qs, $qe)=($qe, $qs) if $qe<$qs; |
| 50 |
my $attr="ID=$qid;Name=$qname;qreg=$qs-$qe|$qlen"; |
| 51 |
$attr.=";Cov=$qcov" if $qcov; |
| 52 |
$attr.=";PID=$t[10];PSIM=$t[11];Score=$t[12]"; |
| 53 |
$attr.=";EValue=$t[19]" if length($t[19])>1; |
| 54 |
my $descr=$rswap ? '' : $t[15]; |
| 55 |
if (length($descr)>2) { |
| 56 |
$attr.=';descr="'.$descr.'"'; |
| 57 |
} |
| 58 |
print join("\t",$gseqid,$t[3],'mRNA', $ex[0]->[0], $ex[-1]->[1], |
| 59 |
$t[11], $strand, '.', $attr)."\n"; |
| 60 |
foreach my $x (@ex) { |
| 61 |
print join("\t",$gseqid,$t[3],'exon', $x->[0], $x->[1], $x->[4], $strand, '.', |
| 62 |
'Parent='.$qid.';qreg='.$x->[2].'-'.$x->[3])."\n"; |
| 63 |
} |
| 64 |
} |
