Parsing GenBank file: get locus tag vs product

Question

Basically, a GenBank file consists on gene entries (announced by 'gene' followed by its corresponding 'CDS' entry (only one per gene) like the two I show here below. I would like to get locus_tag vs product in a tab-delimited two column file. 'gene' and 'CDS' are always preceded and followed by spaces.

A previous question suggested a script.

The problem is that it seems that because 'product' has sometimes '/' character inside its name, its having conflicts with this script, that, as far as I can understand, is using '/' as field separator to store information in an array?

I would like to solve this, either modifying this script or building other one.

perl -nE'
  BEGIN{ ($/, $") = ("CDS", "\t") }
  say "@r[0,1]" if @r= m!/(?:locus_tag|product)="(.+?)"!g and @r>1
' file


 gene            complement(8972..9094)
                 /locus_tag="HAPS_0004"
                 /db_xref="GeneID:7278619"
 CDS             complement(8972..9094)
                 /locus_tag="HAPS_0004"
                 /codon_start=1
                 /transl_table=11
                 /product="hypothetical protein"
                 /protein_id="YP_002474657.1"
                 /db_xref="GI:219870282"
                 /db_xref="GeneID:7278619"
                 /translation="MYYKALAHFLPTLSTMQNILSKSPLSLDFRLLFLAFIDKR"
 gene            68..637
                 /locus_tag="HPNK_00040"
 CDS             68..637
                 /locus_tag="HPNK_00040"
                 /codon_start=1
                 /transl_table=11
                 /product="NinG recombination protein/bacteriophage lambda
                 NinG family protein"
                 /protein_id="CRESA:HPNK_00040"
                 /translation="MIKPKVKKRKCKCCGGEFKSADSFRKWCSAECGVKLAKIAQEKA
                 RQKAIEKRNREERAKIKATRERLKSRSEWLKDAQAIFNEYIRLRDKDEPCISCRRFHQ
                 GQYHAGHYRTVKAMPELRFNEDNVHKQCSACNNHLSGNITEYRINLVRKIGAERVEAL
                 ESYHPPVKWSVEDCKEIIKTYRAKIKELK"

Answer 1

As your sample GenBank file was incomplete, I went online to find a sample file that could be used in an example, and I found this file.

Using this code and the Bio::GenBankParser module, it was parsed guessing what parts of the structure you were after. In this case, "features" that contained both a locus_tag field and a product field.

use strict;
use warnings;
use feature 'say';
use Bio::GenBankParser;

my $file = shift;
my $parser = Bio::GenBankParser->new( file => $file );
while ( my $seq = $parser->next_seq ) {
    my $feat = $seq->{'FEATURES'};
    for my $f (@$feat) {
        my $tag = $f->{'feature'}{'locus_tag'};
        my $prod = $f->{'feature'}{'product'};
        if (defined $tag and defined $prod) {
            say join "\t", $tag, $prod;
        }
    }
}

Usage:

perl script.pl input.txt > output.txt

Output:

MG_001  DNA polymerase III, beta subunit
MG_470  CobQ/CobB/MinD/ParA nucleotide binding domain-containing protein

The output from your one-liner for the same input would be:

MG_001  DNA polymerase III, beta subunit
MG_470  CobQ/CobB/MinD/ParA nucleotide binding
                     domain-containing protein

Assuming of course that you add the /s modifier to the regex to account for multiline entries (which leeduhem pointed out in the comments):

m!/(?:locus_tag|product)="(.+?)"!sg
#                                ^---- this

Answer 2

Having read your duplicated question http://www.biostars.org/p/94164/ (please don't double post like this), here's a minimal Biopython answer:

import sys
from Bio import SeqIO
filename = sys.argv[1] # Takes first command line argument input filename
for record in SeqIO.parse(filename, "genbank"):
    for feature in record.features:
        if feature.type == "CDS":
            locus_tag = feature.qualifiers.get("locus_tag", ["???"])[0]
            product = feature.qualifiers.get("product", ["???"])[0]
            print("%s\t%s" % (locus_tag, product))

With minor changes you can write this out to a file instead.