#!/usr/bin/perl   -w

 

use strict;

use Data::Dumper;

 

my $fastaSuffix =".fasta";

 

@ARGV =qw(-) unless @ARGV;

foreach my $filename (@ARGV)  {

     local *FILE;

open FILE, "<$filename" or die "Couldn't open '$filename': $!";

my $genbank = new GBlite(\*FILE);

while (my $entry= $genbank->nextENTRY)   {

my $ac_number=$entry->AC_number;

my $gene_id=$entry->gene_id;

my $DNA_seq=$entry->DNA_seq;

my $start_stop_cs=$entry->start_stop_cs;

my $aa_seq=$entry->aa_seq;

my $transl_start_site=$entry->transl_start_site;

my $protein_prod_names=$entry->protein_prod_names;

my $chromosomal_location=$entry->chromosomal_location;

 

my $fasta = $ac_number . $fastaSuffix;

local *FASTA;

warn "[working '$fasta']\n";

open FASTA, ">$fasta" or die "Couldn't open '$fasta': $!";

print FASTA ">$ac_number\n", ">$gene_id\n", ">$DNA_seq\n", ">$start_stop_cs\n", ">$aa_seq\n", ">$transl_start_site\n",  ">$protein_prod_names\n", ">$chromosomal_location\n";




 

package GBlite;

use strict;

 

 

sub new  {

my  ($class, $fh) = @_;

if (ref $fh !~ /GLOB/)   {die "GBlite error: new expects a GLOB reference not $fh\n"}                 

my $this =bless{};

$this-> {FH}=$fh;

$this-> {LASTLINE}="";

$this-> {DONE}=0;

return $this;

   }

 

sub netxEntry  {

my ($this)=@_;

$this->fastForward or return 0;

my $FH = $this ->{FH};

 

#Theses are the fields that will be kept

my ($ac_number, $gene_id, $DNA_seq, $start_stop_cs, $aa_seq, $transl_start_site, $protein_prod_names, $chromosomal_location);

 

 

#get ac_number which may be in several lines

my %ac_number;

while (<$FH>) {

   if (/^VERSION/)  {

    $this->{LASTLINE}=$_;

     last;

    }

else   {

$_= ~m/s*(\w+)/;

my @ac_number=split;

foreach my $ac_number(@ac_number)   {

   $ac_number{$ac_number}++;

     }

    }

   }

$ac_number=[keys %ac_number];

 

}