Bio-DB-GFF

 view release on metacpan or  search on metacpan

bin/bp_bulk_load_gff  view on Meta::CPAN

sub insert_sequence {
  my $self = shift;
  my ($id,$offset,$seq) = @_;
  print "$id\t$offset\t$seq\n";
}

package main;
$main::VERSION = '1.7.4';
eval "use Time::HiRes"; undef $@;
my $timer = defined &Time::HiRes::time;

my $bWINDOWS = 0;    # Boolean: is this a MSWindows operating system?
if ($^O =~ /MSWin32/i) {
    $bWINDOWS = 1;
}

my ($DSN,$ADAPTOR,$FORCE,$USER,$PASSWORD,$FASTA,$LOCAL,$MAX_BIN,$GROUP_TAG,$LONG_LIST,$MUNGE,$TMPDIR);

GetOptions ('database:s'    => \$DSN,
	    'adaptor:s'     => \$ADAPTOR,

bin/bp_bulk_load_gff  view on Meta::CPAN

my $FEATURES    = 0;

my %tmpfiles; # keep track of temporary fasta files
my $count;
my $fasta_sequence_id;
my $gff3;
my $current_file; #used to reset GFF3 flag in mix of GFF and GFF3 files

$db->preferred_groups(split (/[,\s]+/,$GROUP_TAG)) if defined $GROUP_TAG;

my $last  = Time::HiRes::time() if $timer;
my $start = $last;

  # avoid hanging on standalone --fasta load
if (!@ARGV) {
    $FH{NULL} = IO::File->new(">$tmpdir/null");
    push @ARGV, "$tmpdir/null";
}

my ($cmap_db);
if ($use_mysqlcmap){

bin/bp_bulk_load_gff  view on Meta::CPAN

    $FH{ FTYPE()  }->print(    join("\t",$ftypeid,$method,$source),"\n"                   ) unless $DONE{"T$ftypeid"}++;

    foreach (@$attributes) {
      my ($key,$value) = @$_;
      my $attributeid = $ATTRIBUTEID{$key}   ||= $ATTRIBUTEID++;
      $FH{ FATTRIBUTE() }->print( join("\t",$attributeid,$key),"\n"                       ) unless $DONE{"A$attributeid"}++;
      $FH{ FATTRIBUTE_TO_FEATURE() }->print( join("\t",$fid,$attributeid,$value),"\n");
    }

    if ( $fid % 1000 == 0) {
      my $now    = Time::HiRes::time() if $timer;
      my $elapsed = $timer ? sprintf(" in %5.2fs",$now - $last) : '';
      $last = $now;
      print STDERR "$fid features parsed$elapsed...";
      print STDERR -t STDOUT && !$ENV{EMACS} ? "\r" : "\n";
    }
  }
}

$FH{FASTA}->close if exists $FH{FASTA};

for my $file (@fasta) {

bin/bp_bulk_load_gff  view on Meta::CPAN

if ($use_pg) {
  $FH{FDATA()                }->print("\\.\n\n");
  $FH{FTYPE()                }->print("\\.\n\n");
  $FH{FGROUP()               }->print("\\.\n\n");
  $FH{FATTRIBUTE()           }->print("\\.\n\n");
  $FH{FATTRIBUTE_TO_FEATURE()}->print("\\.\n\n");
}


$_->close foreach values %FH;
printf STDERR "Total parse time %5.2fs\n",(Time::HiRes::time() - $start) if $timer;
warn "Loading feature data and analyzing tables.  You may see RDBMS messages here...\n";

if ($use_pg){
  warn "Loading feature data.  You may see Postgres comments...\n";

  foreach (@files) {
    my $file = "$tmpdir/$_.$$";

    $AUTH ? system("psql $AUTH -f $file $DBNAME")
          : system('psql','-f', $file, $DBNAME);

bin/bp_bulk_load_gff  view on Meta::CPAN

    my $sql = join ('; ',
		    "lock tables $table write",
		    "delete from $table",
		    "load data $LOCAL infile '$tmpdir/$f.$$' replace into table $table $TERMINATEDBY",
		    "unlock tables");
    my $command = MYSQL . qq[$AUTH -s -e "$sql"];
    $command =~ s/\n/ /g;
    $success &&= system($command) == 0;
    unlink "$tmpdir/$f.$$";
  }
  printf STDERR "Total load time %5.2fs\n",(time() - $start) if $timer;
  print STDERR "done...\n";

  print STDERR "Analyzing/optimizing tables. You will see database messages...\n";
  $start = time();
  my $sql = '';
  for my $f (@files) {
    my $table = function_to_table($f,$ADAPTOR);
    $sql       .= "analyze table $table;";
  }
  my $command = MYSQL . qq[$AUTH -N -s -e "$sql"];

bin/bp_fast_load_gff  view on Meta::CPAN

sub insert_sequence {
  my $self = shift;
  my ($id,$offset,$seq) = @_;
  print join "\t",$id,$offset,$seq,"\n";
}

package main;
$main::VERSION = '1.7.4';
eval "use Time::HiRes"; undef $@;
my $timer = defined &Time::HiRes::time;

my ($DSN,$CREATE,$USER,$PASSWORD,$FASTA,$FAILED,$LOCAL,%PID,$MAX_BIN,$GROUP_TAG,$LONG_LIST,$MUNGE,$TMPDIR,$SUMMARY_STATS);

if ($DO_FAST) {
  $SIG{CHLD} = sub {
    while ((my $child = waitpid(-1,&WNOHANG)) > 0) {
      delete $PID{$child} or next;
      $FAILED++ if $? != 0;
    }
  }

bin/bp_fast_load_gff  view on Meta::CPAN

  print STDERR "opening load file for writing...";
  $FH{$_} = IO::File->new($file,'>') or die $_,": $!";
  print STDERR "ok\n";
  $FH{$_}->autoflush;
}

print STDERR "Fast loading enabled\n"    if $DO_FAST;

my ($count,$gff3,$last,$start,$beginning,$current_file);

$last  = Time::HiRes::time() if $timer;
$beginning = $start = $last;

# avoid hanging on standalone --fasta load
if (!@ARGV) {
    $FH{NULL} = IO::File->new(">$tmpdir/null");
    push @ARGV, "$tmpdir/null";
}

while (<>) {

bin/bp_fast_load_gff  view on Meta::CPAN

    $FH{ FTYPE()  }->print(    join("\t",$ftypeid,$method,$source),"\n"                   ) unless $DONE{"ftype$;$ftypeid"}++;

    foreach (@$attributes) {
      my ($key,$value) = @$_;
      my $attributeid = $ATTRIBUTEID{lc $key}   ||= $ATTRIBUTEID++;
      $FH{ FATTRIBUTE() }->print( join("\t",$attributeid,$key),"\n"                       ) unless $DONE{"fattribute$;$attributeid"}++;
      $FH{ FATTRIBUTE_TO_FEATURE() }->print( join("\t",$fid,$attributeid,$value),"\n");
    }

    if ( $FEATURES % 1000 == 0) {
      my $now    = Time::HiRes::time() if $timer;
      my $elapsed = $timer ? sprintf(" in %5.2fs",$now - $last) : '';
      $last = $now;
      print STDERR "$fid features parsed$elapsed...";
      print STDERR -t STDOUT && !$ENV{EMACS} ? "\r" : "\n";
    }
  }
}

$FH{FASTA}->close if exists $FH{FASTA};

printf STDERR "Feature load time %5.2fs\n",(Time::HiRes::time() - $start) if $timer;
$start = time();

for my $fasta (@fasta) {
  warn "Loading fasta ",(-d $fasta?"directory":"file"), " $fasta\n";
  my $old = select($FH{FDNA()});
  my $loaded = $db->load_fasta($fasta);
  warn "$fasta: $loaded records loaded\n";
  select $old;
}

printf STDERR "Fasta load time %5.2fs\n",(Time::HiRes::time() - $start) if $timer;
$start = time();

my $success = 1;
if ($DO_FAST) {
  warn "Indexing and analyzing tables.  This may take some time (you may see database messages during the process)...\n";
}

$_->close foreach values %FH;

if (!$DO_FAST) {

bin/bp_fast_load_gff  view on Meta::CPAN

}

# wait for children
while (%PID) {
  sleep;
}
$success &&= !$FAILED;

cleanup();

printf STDERR "Total parse & load time %5.2fs\n",(Time::HiRes::time() - $beginning) if $timer;

if ($success) {
  print "SUCCESS: $FEATURES features successfully loaded\n";
  exit 0;
} else {
  print "FAILURE: Please see standard error for details\n";
  exit -1;
}

if ($SUMMARY_STATS) {