| 1 | #!/usr/bin/perl -w |
|---|
| 2 | |
|---|
| 3 | use strict; |
|---|
| 4 | use warnings; |
|---|
| 5 | |
|---|
| 6 | my @columns = (); |
|---|
| 7 | my $del = ""; |
|---|
| 8 | my @in = (); |
|---|
| 9 | my @out = (); |
|---|
| 10 | my $command = ""; |
|---|
| 11 | my $field = 0; |
|---|
| 12 | |
|---|
| 13 | # a wrapper for cut for use in galaxy |
|---|
| 14 | # cutWrapper.pl [filename] [columns] [delim] [output] |
|---|
| 15 | |
|---|
| 16 | die "Check arguments\n" unless @ARGV == 4; |
|---|
| 17 | |
|---|
| 18 | $ARGV[1] =~ s/\s+//g; |
|---|
| 19 | foreach ( split /,/, $ARGV[1] ) { |
|---|
| 20 | if (m/^c\d{1,}$/i) { |
|---|
| 21 | push (@columns, $_); |
|---|
| 22 | $columns[@columns-1] =~s/c//ig; |
|---|
| 23 | } |
|---|
| 24 | } |
|---|
| 25 | |
|---|
| 26 | die "No columns specified, columns are not preceded with 'c', or commas are not used to separate column numbers: $ARGV[1]\n" if @columns == 0; |
|---|
| 27 | |
|---|
| 28 | my $column_delimiters_href = { |
|---|
| 29 | 'T' => q{\t}, |
|---|
| 30 | 'C' => ",", |
|---|
| 31 | 'D' => "-", |
|---|
| 32 | 'U' => "_", |
|---|
| 33 | 'P' => q{\|}, |
|---|
| 34 | 'Dt' => q{\.}, |
|---|
| 35 | 'Sp' => q{\s+} |
|---|
| 36 | }; |
|---|
| 37 | |
|---|
| 38 | $del = $column_delimiters_href->{$ARGV[2]}; |
|---|
| 39 | |
|---|
| 40 | open (OUT, ">$ARGV[3]") or die "Cannot create $ARGV[2]:$!\n"; |
|---|
| 41 | open (IN, "<$ARGV[0]") or die "Cannot open $ARGV[0]:$!\n"; |
|---|
| 42 | while (<IN>) { |
|---|
| 43 | chop; |
|---|
| 44 | @in = split /$del/; |
|---|
| 45 | foreach $field (@columns) { |
|---|
| 46 | if (defined($in[$field-1])) { |
|---|
| 47 | push(@out, $in[$field-1]); |
|---|
| 48 | } else { |
|---|
| 49 | push(@out, "."); |
|---|
| 50 | } |
|---|
| 51 | } |
|---|
| 52 | print OUT join("\t",@out), "\n"; |
|---|
| 53 | @out = (); |
|---|
| 54 | } |
|---|
| 55 | close IN; |
|---|
| 56 | |
|---|
| 57 | close OUT; |
|---|
| 58 | |
|---|