scripts/DB/bp_bioflat_index.pl

   1 #!/usr/bin/perl
   2
   3
   4 =head1 NAME
   5
   6 bp_bioflat_index.pl - index sequence files using Bio::DB::Flat
   7
   8 =head1 DESCRIPTION
   9
  10  Create or update a biological sequence database indexed with the
  11  Bio::DB::Flat indexing scheme.  The arguments are a list of flat files
  12  containing the sequence information to be indexed.
  13
  14 =head1 USAGE
  15
  16  bp_bioflat_index.pl <options> file1 file2 file3...
  17
  18  Options:
  19
  20      --create              Create or reinitialize the index.  If not specified,
  21                            the index must already exist.
  22
  23      --format   <format>   The format of the sequence files.  Must be one
  24                            of "genbank", "swissprot", "embl" or "fasta".
  25
  26      --location <path>     Path to the directory in which the index files
  27                            are stored.
  28
  29      --dbname <name>       The symbolic name of the database to be created.
  30
  31      --indextype <type>    Type of index to create.  Either "bdb" or "flat".
  32                            "binarysearch" is the same as "flat".
  33
  34 Options can be abbreviated.  For example, use -i for --indextype.
  35
  36 The following environment variables will be used as defaults if the
  37 corresponding options are not provided:
  38
  39      OBDA_FORMAT      format of sequence file
  40      OBDA_LOCATION    path to directory in which index files are stored
  41      OBDA_DBNAME      name of database
  42      OBDA_INDEX       type of index to create
  43
  44 =cut
  45
  46 use strict;
  47 use warnings;
  48 use Bio::Root::Root;
  49 use Bio::Root::IO;
  50 use Bio::DB::Flat;
  51 use Getopt::Long;
  52 use File::Path qw(mkpath rmtree);
  53
  54 my ($CREATE,$FORMAT,$LOCATION,$DBNAME,$INDEXTYPE);
  55
  56 GetOptions( 'create'      => \$CREATE,
  57                                 'format:s'    => \$FORMAT,
  58                                 'location:s'  => \$LOCATION,
  59                                 'dbname:s'    => \$DBNAME,
  60                                 'indextype:s' => \$INDEXTYPE );
  61
  62 $FORMAT    = $ENV{OBDA_FORMAT}    unless defined $FORMAT;
  63 $LOCATION  = $ENV{OBDA_LOCATION}  unless defined $LOCATION;
  64 $DBNAME    = $ENV{OBDA_DBNAME}    unless defined $DBNAME;
  65 $INDEXTYPE = $ENV{OBDA_INDEXTYPE} unless defined $INDEXTYPE;
  66
  67 my $root = 'Bio::Root::Root';
  68 my $io   = 'Bio::Root::IO';
  69
  70 # confirm that database directory is there
  71 defined $LOCATION or
  72   $root->throw("please provide a base directory with the --location option");
  73
  74 -d $LOCATION or
  75   $root->throw("$LOCATION is not a valid directory; use --create to create a new index");
  76
  77 defined $DBNAME or
  78   $root->throw("please provide a database name with the --dbname option");
  79
  80 defined $FORMAT or
  81   $root->throw("please specify the format for the input files with the --format option");
  82
  83 unless (defined $INDEXTYPE) {
  84         $INDEXTYPE = 'flat';
  85         $root->warn('setting index type to "flat", use the --indextype option to override');
  86 }
  87
  88 # Confirm that database is there and that --create flag is sensible.
  89 my $path = $io->catfile($LOCATION,$DBNAME,'config.dat');
  90 if (-e $path) {
  91   if ($CREATE) {
  92     $root->warn("existing index detected; deleting.");
  93     rmtree($io->catfile($LOCATION,$DBNAME),1,1);
  94   } else {
  95     $root->warn("existing index detected; ignoring --indextype and --format options.");
  96     undef $INDEXTYPE;
  97   }
  98 }
  99 elsif (!$CREATE) {
 100   $root->throw("Cannot find database config file at location $path; use --create to create a new index");
 101 }
 102
 103 # open for writing/updating
 104 my $db = Bio::DB::Flat->new(-directory  => $LOCATION,
 105                                                                          -dbname     => $DBNAME,
 106                                                                          $INDEXTYPE ? (
 107                                                                                                                 -index      => $INDEXTYPE
 108                                                                                                           )
 109                                                                          : (),
 110                                                                          -write_flag => 1,
 111                                                                          -format     => $FORMAT) or
 112   $root->throw("can't create Bio::DB::Flat object");
 113
 114 my $entries = $db->build_index(@ARGV);
 115
 116 print STDERR "(Re)indexed $entries entries.\n ";
 117
 118 __END__