G
Skyline
| Globals (from use vars definitions) |
| $AUTOLOAD |
| @INC |
| $VERSION = '1.42' |
| Privates (from my definitions) |
| $loaded = 0 |
| $outfile = '' |
| $infile = '' |
use G::Skyline;
@ISA = (G::Skyline);
|
Inherits all necessary classes.
Intended for internal use only.
|
| BEGIN | | Code |
| DESTROY | No description | Code |
| bioperl | No description | Code |
| filepath | No description | Code |
| interpret_format | No description | Code |
| loaded_msg | No description | Code |
| multi_locus | No description | Code |
| new | No description | Code |
| output | No description | Code |
Methods description
Methods code
BEGIN
{ eval "use Bio::SeqIO;";
if($@){ warn "$@" };
eval "use Bio::DB::GenBank;";
if($@){ warn "$@" };
eval "use Bio::DB::RefSeq;";
if($@){ warn "$@" };}
sub DESTROY
{ my $self = shift;
undef %{$self};
unlink($outfile);
$self->close_gb();}
sub bioperl
{ my $this = shift;
my $outfile = "/tmp/" . time . '.gbk';
$this->output($outfile);
my $in = Bio::SeqIO->new(-format => "genbank", -file =>$outfile);
my $bp = $in->next_seq();
## unlink($outfile);
return $bp;}
sub filepath
{ my $this = shift;
return $infile;}
sub interpret_format
{ my $filename = shift;
if (ref $filename =~ /Bio::Seq/){
return 'bioperl';
}elsif (lc($filename) =~ /\.(gb|gbk|gbank|genbank)$/i){
return 'genbank';
}elsif (lc($filename) =~ /\.(fasta|fast|seq|fst|fa|fsa|nt|aa)/i){
return 'fasta';
}elsif ($filename =~ /^NC_\d+$/i){
return 'RefSeq';
}elsif ($filename =~ /^(?:[A-Z]|[A-Z][A-Z])\d+$/i){
return 'net GenBank';
}else{
my $format = Bio::SeqIO->_guess_format($filename);
if (length $format){
return $format;
}else{
warn("Unknown file format. Interpreting $filename as GenBank...\n");
return 'genbank';
}
}}
sub loaded_msg
{ my $this = shift;
$loaded ++;
return if ($loaded > 1);
my $print =
qq(
__/__/__/__/__/__/__/__/__/__/__/__/__/
G-language Genome Analysis Environment
Version: $VERSION
License: GPL
Copyright (C) 2001-2002
G-language Project
Institute for Advanced Biosciences,
Keio University, JAPAN
http://www.g-language.org/
__/__/__/__/__/__/__/__/__/__/__/__/__/\n
);
&msg_error($print);}
sub multi_locus
{ my $this = shift;
my $gb = shift;
my $lng;
my $i = 1;
my $f = 1;
my $c = 1;
do{
my $F = 1;
my $C = 1;
$lng = length($this->{"SEQ"});
$this->{"LOCUS$i"} = $gb->{"LOCUS"};
$this->{"HEADER$i"} = $gb->{"HEADER"};
$this->{"COMMENT$i"} = $gb->{"COMMENT"};
while(defined (%{$gb->{"FEATURE$F"}})){
$this->{"FEATURE$f"} = $gb->{"FEATURE$F"};
$this->{"FEATURE$f"}->{"start"} = $gb->{"FEATURE$F"}->{"start"} + $lng;
$this->{"FEATURE$f"}->{"end"} = $gb->{"FEATURE$F"}->{"end"} + $lng;
$this->{"FEATURE$f"}->{"locus"} = $i;
if($gb->{"FEATURE$F"}->{"type"} eq "CDS"){
$this->{"CDS$c"} = $gb->{"CDS$C"};
$this->{"CDS$c"}->{"start"} = $gb->{"CDS$C"}->{"start"} + $lng;
$this->{"CDS$c"}->{"end"} = $gb->{"CDS$C"}->{"end"} + $lng;
$this->{"CDS$c"}->{"feature"} = $f;
$this->{"CDS$c"}->{"locus"} = $i;
if(defined $gb->{"CDS$C"}->{"join"}){
my @join = split(/\,/,$gb->{"CDS$C"}->{"join"});
my @num = ();
my @new_join = ();
foreach(@join){
if(tr/c/c/){
@num = split(/\.\./,$_);
push (@new_join, sprintf ("c%d\.\.%d", $num[0] + $lng, $num[1] + $lng));
} else {
@num = split(/\.\./,$_);
push (@new_join, sprintf ("%d\.\.%d", $num[0] + $lng, $num[1] + $lng));
}
}
$this->{"CDS$c"}->{join} = join(',', @new_join);
}
$this->{"FEATURE$f"}->{"cds"} = $c;
$c++;
$C++;
}
$f++;
$F++;
}
$this->{"SEQ"} .= $gb->{"SEQ"};
$i++;
}while($gb->next_locus("no msg"));
$this->{"FEATURE0"}->{"type"} = "source";
$this->{"FEATURE0"}->{"start"} = 1;
$this->{"FEATURE0"}->{"end"} = length($this->{"SEQ"});
$this->{"LOCUS"} = $this->{"LOCUS1"};
$this->{"HEADER"} = $this->{"HEADER1"};
$this->{"COMMENT"} = $this->{"COMMENT1"};
$this->{"CDS0"}->{dummy} = 1;}
sub new
{ my $pkg = shift;
my $filename = shift;
my @options = @_;
my $this = {};
my $tmp = {};
$infile = $filename;
bless $this;
my $no_msg = 0;
my $without_annotation = 0;
my $multiple_locus = 0;
my $long_sequence = 0;
my $bioperl = 0;
my $netgbk = 0;
my $longest_ORF_annotation = 0;
my $glimmer_annotation = 0;
my $format = '';
my $locus_msg = '';
return $this if ($filename eq 'blessed');
foreach my $opt_tmp (@options){
my $opt = lc($opt_tmp);
if ($opt =~ /no msg/){
$no_msg = 1;
}elsif ($opt =~ /without annotation/){
$without_annotation = 1;
}elsif ($opt =~ /multiple locus/){
$multiple_locus = 1;
}elsif ($opt =~ /long sequence/){
$long_sequence = 1;
}elsif ($opt =~ /bioperl/){
$bioperl = 1;
}elsif ($opt =~ /longest ORF annotation/){
$longest_ORF_annotation = 1;
}elsif ($opt =~ /glimmer annotation/){
$glimmer_annotation = 1;
}elsif ($opt =~ /net GenBank/){
$netgbk = 1;
}elsif ( lc($opt) eq 'fasta' || lc($opt) eq 'embl'
|| lc($opt) eq 'swiss' || lc($opt) eq 'scf'
|| lc($opt) eq 'pir' || lc($opt) eq 'gcg'
|| lc($opt) eq 'raw' || lc($opt) eq 'ace'
|| lc($opt) eq 'game' || lc($opt) eq 'phd'
|| lc($opt) eq 'qual' || lc($opt) eq 'bsml'
|| lc($opt) eq 'genbank'){
$format = lc($opt);
}else{
unless (lc($opt) eq 'net genbank'){
warn("Warning: Unknown Option $opt at\" new G\"\n");
}
}
}
$this->loaded_msg() unless ($no_msg);
$format = interpret_format($filename) unless(length $format);
if ($bioperl || $format eq 'bioperl'){
G::IO::Bioperl::convert($filename, $this);
$this->seq_info() unless ($no_msg);
}elsif ($format eq 'RefSeq'){
my $db = new Bio::DB::RefSeq();
my $bp;
eval {
$bp = $db->get_Seq_by_id($filename);
};
die("$@ Could not retrieve $filename\n") if $@;
G::IO::Bioperl::convert($bp, $this);
$this->seq_info() unless($no_msg);
}elsif ($format eq 'net GenBank'){
my $db = new Bio::DB::GenBank();
my $bp;
eval {
$bp = $db->get_Seq_by_acc($filename);
};
die("$@ Could not retrieve $filename\n") if $@;
G::IO::Bioperl::convert($bp, $this);
$this->seq_info() unless($no_msg);
}else{
if ($format eq 'embl'){
$this = new G::IO::EmblI;
}elsif ($format eq 'fasta'){
$this = new G::IO::FastaI;
}elsif ($format ne 'genbank'){
$outfile = '/tmp/' . time . '.gbk';
my $in = Bio::SeqIO->newFh(-format => $format, -file => $filename);
my $out = Bio::SeqIO->newFh(-format => "GenBank",
-file => '>' . $outfile);
print $out $_ while <$in>;
$filename = $outfile;
$locus_msg = 'no msg';
$format = 'genbank';
}
*GENBANK = $this->open_gb($filename);
if ($multiple_locus){
$tmp = new G::Skyline($filename, "no msg", $format);
$this->multi_locus($tmp);
$this->seq_info() unless($no_msg);
}else{
if ($without_annotation){
$this->goto_origin();
}else{
$this->read_locus($locus_msg);
$this->read_header();
$this->read_features();
}
if ($long_sequence){
$this->{origin} = tell *GENBANK;
}else{
$this->getnucs();
$this->seq_info() unless ($no_msg);
}
}
}
if ($longest_ORF_annotation){
my $new = new G::Skyline;
bless $new;
annotate_with_LORF($new, $this);
return $new;
}elsif ($glimmer_annotation){
my $new = new G::Skyline;
bless $new;
open(FASTA, '>/tmp/out.fasta') || die ($!);
printf FASTA ">%s\n%s\n", $this->{LOCUS}->{id}, $this->{SEQ};
close(FASTA);
run_glimmer($this, '/tmp/out.fasta');
annotate_with_glimmer($new, '/tmp/out.fasta');
unlink('/tmp/out.fasta');
return $new;
}
return $this;}
sub output
{ my $gb = shift;
my $file = shift;
my $option = shift;
$option = interpret_format($file) unless(length $option);
if (lc($option) eq 'genbank'){
$gb->make_gb($file);
}elsif(length $option){
my $outfile = '/tmp/' . time . '.gbk';
$gb->make_gb($outfile);
my $in = Bio::SeqIO->newFh(-format => "genbank", -file => $outfile);
my $out = Bio::SeqIO->newFh(-format => "$option",
-file => '>' . $file);
print $out $_ while <$in>;
}else{
&msg_error("G::output - Unknown format to output.");
}}
General documentation