| 1 |
gpertea |
23 |
#!/usr/bin/perl |
| 2 |
|
|
use strict; |
| 3 |
|
|
use Getopt::Std; |
| 4 |
|
|
#use FindBin;use lib $FindBin::Bin; |
| 5 |
|
|
|
| 6 |
|
|
my $usage = q/Usage: |
| 7 |
|
|
|
| 8 |
|
|
cat tab_delim_file | trcol.pl [-c <col#>] [-G] <map_file> |
| 9 |
|
|
|
| 10 |
|
|
Replaces (translates) column <col#> (default 1) from tab delimited |
| 11 |
|
|
input with the corresponding string according to the word |
| 12 |
|
|
translation table in <map_file>, which is a simple space delimited, |
| 13 |
|
|
two column mapping table of the format: |
| 14 |
|
|
|
| 15 |
|
|
<string_to_find> <replacement> |
| 16 |
|
|
|
| 17 |
|
|
Use -G option if the input is GFF3 format and to only output the |
| 18 |
|
|
replaced column and the value of the ID field for the mRNA |
| 19 |
|
|
features only. |
| 20 |
|
|
|
| 21 |
|
|
|
| 22 |
|
|
/; |
| 23 |
|
|
umask 0002; |
| 24 |
|
|
getopts('Gc:') || die($usage."\n"); |
| 25 |
|
|
my $replcol=$Getopt::Std::opt_c || 1; #column to replace |
| 26 |
|
|
my $gffmap=$Getopt::Std::opt_G; |
| 27 |
|
|
$replcol--; |
| 28 |
|
|
die("$usage\n") if $replcol<0; |
| 29 |
|
|
my $xtable=$ARGV[0]; |
| 30 |
|
|
die ("$usage\n") unless $xtable && -f $xtable; |
| 31 |
|
|
my %xw; |
| 32 |
|
|
open(XTBL, $xtable) || die ("Error opening file $xtable!\n"); |
| 33 |
|
|
while (<XTBL>) { |
| 34 |
|
|
chomp; |
| 35 |
|
|
next unless $_; |
| 36 |
|
|
my @t=split; |
| 37 |
|
|
$xw{$t[0]}=$t[1]; |
| 38 |
|
|
} |
| 39 |
|
|
while (<STDIN>) { |
| 40 |
|
|
chomp; |
| 41 |
|
|
my @t=split(/\t/); |
| 42 |
|
|
my $tx=$xw{$t[$replcol]}; |
| 43 |
|
|
$t[$replcol]=$tx if $tx; |
| 44 |
|
|
if ($gffmap) { |
| 45 |
|
|
next unless m/ID=([^;]+)/; |
| 46 |
|
|
my $id=$1; |
| 47 |
|
|
$id=~s/\.\w+\d$//; |
| 48 |
|
|
print $t[$replcol]."\t$id\n"; |
| 49 |
|
|
} |
| 50 |
|
|
else { |
| 51 |
|
|
print join("\t",@t)."\n"; |
| 52 |
|
|
} |
| 53 |
|
|
} |
