#!/usr/bin/perl
##---------------------------------------------------------------------------##
##  File:
##      @(#) RepbaseEMBL.pm
##  Author:
##      Robert M. Hubley   rhubley@systemsbiology.org
##  Description:
##      A simple EMBL object.  This is a simple datastructure
##      for encapsulating the EMBL formatted sequence data and
##      annotations.  **THIS IS NOT A COMPLETE IMPLEMENTATION OF
##      THE EMBL FORMAT** but rather restricted to the features
##      currently used by Repbase. Two methods provide the ability
##      to read/write Repbase EMBL formatted files.
##
#******************************************************************************
#* Copyright (C) Institute for Systems Biology 2003-2004 Developed by
#* Robert Hubley.
#*
#* This work is licensed under the Open Source License v2.1.  To view a copy
#* of this license, visit http://www.opensource.org/licenses/osl-2.1.php or
#* see the license.txt file contained in this distribution.
#*
#******************************************************************************
# Implementation Details:
#
#    The internal data storage is defined below.  Data accessor
#    methods have not been built ( but should be before external
#    access is allowed ).  Currently the file's records ( defined
#    below ) are stored in an array:
#
#    bless( {'records'} => [ {#EMBL Record}, {#EMBL Record}, .. ]},
#          'RepbaseEMBL' );
#
#    Repbase/EMBL Record                                 EMBL Tag
#    ------------------------------------------------    --------
#    { 'id' => blah,                                     ID
#      'dataclass' => blah,                              ID
#      'molecule' => blah,                               ID
#      'divisions' => [ blah, blah, blah ],              ID
#      'accessions' => [ blah, blah, blah ],             AC
#      'dateCreated' => blah,                            DT
#      'dateUpdated' => blah,                            DT
#      'comments' => [ blah, blah, blah, blah ],         CC
#      'keywords' => [ blah, blah, blah, blah ],         KW
#      'description' => blah,                            DE
#      'speciesName' => blah,                            OS
#      'commonSpeciesName' => blah,                      OS
#      'classification' => [ blah, blah, blah, blah ],   OC
#      'references' => [ { 'number' = > blah,            RN
#                          'comment' => blah,            RC
#                          'position' => blah,           RP
#                          'xref' => blah,               RX
#                          'authors' => blah,            RA
#                          'title' => blah,              RT
#                          'location' => blah } ],       RL
#       'xref' => blah,                                  DR
#       'length' => blah,                                ID
#       'composition' => { 'A' => blah,                  SQ
#                          'C' => blah,                  SQ
#                          'G' => blah,                  SQ
#                          'T' => blah,                  SQ
#                          'other' => 'blah' },          SQ
#       'sequence' => blah };                            SQ
#
#******************************************************************************
#

=head1 NAME

RepbaseEMBL


=head1 SYNOPSIS

use RepbaseEMBL

Usage:

    $DB = RepbaseEMBL->new( fileName => "humrep.ref" );

=head1 DESCRIPTION

A simple EMBL object.  This is a simple datastructure
for encapsulating the EMBL formatted sequence data and
annotations.  **THIS IS NOT A COMPLETE IMPLEMENTATION OF
THE EMBL FORMAT** but rather restricted to the features
currently used by Repbase. Two methods provide the ability
to read/write Repbase EMBL formatted files.

Caveats:
   - Currently no attempt is made to store the ordering
     of tag groups.  The output ordering is fixed and
     may not reflect the original input ordering.  Ie.
     it appears there is a convention of using a "CC"
     tag following an "OS" tag to store the common name
     of a speciess:

         OS   Anopheles gambiae
         CC   African malaria mosquito.

     The EMBL Standard requires that this information be
     placed on the "OS" line and surrounded by parantheses.
     The reader does not currently differentiate between
     "CC" locations in a record.  Therefore this will be
     reformatted upon output as:

         OS   Anopheles gambiae
         OC   ..
         ..
         XX
         CC   African malaria mosquito.

   - The line breaks/spacing in comment blocks are potentially
     important.  This object attempts to keep save the line
     break and spacing information for several types of
     data including comment tags.  It will be the responsibility
     of the data user to verify line width and reformat paragraphs
     as the data is manipulated.

=head1 INSTANCE METHODS

=cut

package RepbaseEMBL;
use strict;
use PubRef;
use RepbaseRecord;
use Carp;
use Data::Dumper;
use Text::Wrap;
use vars qw(@ISA @EXPORT @EXPORT_OK %EXPORT_TAGS $VERSION);

require Exporter;

@ISA = qw();

@EXPORT = qw();

@EXPORT_OK = qw();

%EXPORT_TAGS = ( all => [ @EXPORT_OK ] );

#
# Version
#
my $VERSION = "0.1";
my $CLASS   = "RepbaseEMBL";

#
# EMBL Tags used by Repbase
#
my %validTagBits = (
                     'ID' => 1,
                     'CC' => 2,
                     'AC' => 4,
                     'DT' => 8,
                     'DE' => 16,
                     'KW' => 32,
                     'OS' => 64,
                     'OC' => 128,
                     'RN' => 256,
                     'RP' => 512,
                     'RC' => 1024,
                     'RA' => 2048,
                     'RT' => 4096,
                     'RL' => 8192,
                     'DR' => 16384,
                     'SQ' => 32768,
                     'RX' => 65536,
                     'FT' => 131072,
                     'RG' => 262144,
                     'FH' => 524288,
                     'NM' => 1048576
);

##-------------------------------------------------------------------------##
## Constructor
##-------------------------------------------------------------------------##
sub new {
  my $class = shift;

  my %nameValuePairs = @_;

  # Create ourself as a hash
  my $this = {};

  $this->{'records'} = [];

  # Bless this hash in the name of the father, the son...
  bless $this, $class;

  if ( %nameValuePairs ) {
    my %validKeys = ( fileName => 1 );
    while ( my ( $name, $value ) = each( %nameValuePairs ) ) {
      unless ( exists $validKeys{$name} ) {
        croak "$CLASS::new: Error $name is not a valid "
            . "parameter to the constructor!\n\n"
            . "Usage: \$DB = $CLASS->new();\n"
            . "       \$DB = $CLASS->new( fileName=>\"humrep.ref\" );\n";
      }
      $this->{$name} = $value;
    }
  }

  if ( exists $this->{'fileName'} ) {
    _parseFromFile( $this, $this->{'fileName'} );
  }

  return $this;
}

##-------------------------------------------------------------------------##
## Get and Set Methods
##-------------------------------------------------------------------------##

##-------------------------------------------------------------------------##
## General Object Methods
##-------------------------------------------------------------------------##

##-------------------------------------------------------------------------##

=head2 Use: $DB->writeEMBLFile( $fileName );

  Write out this object as an EMBL file;

=cut

##-------------------------------------------------------------------------##
sub writeEMBLFile {
  my $this     = shift;
  my $fileName = shift;

  open OUT, ">$fileName"
      or die "$CLASS::writeEMBLFile() Unable to open "
      . "file $fileName for writing: $!\n";

  foreach my $record ( @{ $this->{'records'} } ) {
    print OUT "" . &toEMBLString( $record );
  }
  close OUT;

}

##-------------------------------------------------------------------------##

=head2 Use: my $string = toEMBLString( $recordRef );

  Convert a single EMBL record back to EMBL format.
  NOTE: This is a static method.

=cut

##-------------------------------------------------------------------------##
sub toEMBLString {
  my $recRef = shift;    # EMBL Data Structure

  my $outStr   = toEMBLHeader( $recRef );
  my $seq      = $recRef->getSequence();
  my $seqIndex = 0;
  while ( ( $seq =~ s/^(.{60})// ) ) {
    $seqIndex += 60;
    my $lineSeq = $1;
    $lineSeq =~ s/(\w{10})/$1 /g;
    $outStr .= "     " . $lineSeq . "  $seqIndex\n";
  }
  $seqIndex += length( $seq );
  $seq =~ s/(\w{10})/$1 /g;
  my $numPadding = 68 - length( $seq );
  $outStr .= "     " . $seq . ( " " x $numPadding ) . $seqIndex . "\n";
  $outStr .= "//\n";

  return $outStr;

}

sub toEMBLHeader {
  my $recRef = shift;    # EMBL Data Structure

  my $tmpStr = "";
  my $outStr = "";
  $outStr = "ID   "
      . $recRef->getId() . " "
      . $recRef->getDataclass() . "; "
      . $recRef->getMolecule() . "; "
      . $recRef->getDivisionsElement( 0 ) . "; "
      . $recRef->getLength()
      . " BP.\n";
  $outStr .= "XX\n";

  #
  my $accessionStr = "";
  if ( $recRef->getAccessionsCount() > 0 ) {
    $accessionStr = join( "; ", $recRef->getAccessionsArray() );
  }
  ## TODO Need a routine to break lines here
  $accessionStr =~ s/^/AC   /g;
  $outStr .= $accessionStr . "\n";
  $outStr .= "XX\n";

  #
  if ( $recRef->getDateCreated() ne "" ) {
    $outStr .= "DT   " . $recRef->getDateCreated() . "\n";
  }
  if ( $recRef->getDateUpdated() ne "" ) {
    $outStr .= "DT   " . $recRef->getDateUpdated() . "\n";
  }
  $outStr .= "XX\n";

  #
  if ( $recRef->getDescription() ne "" ) {
    $tmpStr = $recRef->getDescription();
    $tmpStr =~ s/^/DE   /gm;
    $outStr .= $tmpStr;
    $outStr .= "XX\n";
  }

  #
  # Keywords
  #
  if ( defined $recRef->getKeywordsArray() ) {
    $tmpStr = join( "; ", $recRef->getKeywordsArray() );
    $outStr .= "KW   $tmpStr.\n";
    $outStr .= "XX\n";
  }

  #
  # Organism Name
  #
  $outStr .= "OS   " . $recRef->getSpeciesName();
  if ( $recRef->getCommonSpeciesName() ne "" ) {
    $outStr .= "(" . $recRef->getCommonSpeciesName() . ")";
  }
  $outStr .= "\n";

  #
  # Organism Classification
  #
  if ( $recRef->getClassificationCount() > 0 ) {
    $outStr .= "OC   ";
    $tmpStr = join( "; ", $recRef->getClassificationArray() ) . ".";

    # TODO Break up lines!

    $outStr .= $tmpStr . "\n";
    $outStr .= "XX\n";
  }

  #
  # References
  #
  foreach my $reference ( $recRef->getReferencesArray() ) {
    $outStr .= "RN   " . $reference->getNumber();
    if ( $reference->getPosition() ne "" ) {
      $tmpStr = $reference->getPosition();
      $tmpStr =~ s/^/RP   /gm;
      $outStr .= $tmpStr;
    }
    if ( $reference->getComment ne "" ) {
      $tmpStr = $reference->getComment();
      $tmpStr =~ s/^/RC   /gm;
      $outStr .= $tmpStr;
    }
    if ( $reference->getAuthors() ne "" ) {
      $tmpStr = $reference->getAuthors();
      $tmpStr =~ s/^/RA   /gm;
      $outStr .= $tmpStr;
    }
    if ( $reference->getTitle() ne "" ) {
      $tmpStr = $reference->getTitle();
      $tmpStr =~ s/^/RT   /gm;
      $outStr .= $tmpStr;
    }
    if ( $reference->getLocation() ne "" ) {
      $tmpStr = $reference->getLocation();
      $tmpStr =~ s/^/RL   /gm;
      $outStr .= $tmpStr;
    }
    if ( $reference->getXref() ne "" ) {
      $tmpStr = $reference->getXref();
      $tmpStr =~ s/^/RX   /gm;
      $outStr .= $tmpStr;
    }
    $outStr .= "XX\n";
  }

  #
  # Database references
  #
  if ( $recRef->getXref() ne "" ) {
    $tmpStr = $recRef->getXref();
    $tmpStr =~ s/^/DR   /gm;
    $outStr .= $tmpStr;
    $outStr .= "XX\n";
  }

  #
  # Comments
  #
  foreach my $comment ( $recRef->getCommentsArray() ) {
    $comment =~ s/^/CC   /gm;
    $outStr .= $comment;
    $outStr .= "XX\n";
  }

  #
  # FT Lines
  #
  my $ftLines = $recRef->getFTLines();
  if ( $ftLines ) {
    $outStr .= $ftLines;
  }

  #
  # Sequence
  #
  $outStr .= "SQ   Sequence " . $recRef->getLength() . " BP;";
  foreach my $base ( "A", "C", "G", "T", "other" ) {
    if ( defined $recRef->getCompositionElement( $base ) ) {
      $outStr .= " " . $recRef->getCompositionElement( $base ) . " $base;";
    }
  }
  $outStr .= "\n";

  return ( $outStr );

}

sub getRecordCount {
  my $this = shift;

  return ( $#{ $this->{'records'} } + 1 );

}

sub getRecord {
  my $this  = shift;
  my $index = shift;

  return if ( $index < 0 || $index > $#{ $this->{'records'} } );
  return $this->{'records'}->[ $index ];

}

sub getRelease {
  my $this = shift;

  return $this->{'release'};
}

##-------------------------------------------------------------------------##
## Private Object Methods
##-------------------------------------------------------------------------##

##-------------------------------------------------------------------------##
## Use: my _parseFromFile( $obj, $filename );
##
##   Parse a FASTA output file into this datastructure.
##
##-------------------------------------------------------------------------##
sub _parseFromFile {
  my $this     = shift;
  my $filename = shift;

  open IN, "<$filename"
      || die "$CLASS::_parseFromFile() Unable to open "
      . "Repbase/EMBL file $filename: $!\n";

  my $prevTag          = "";
  my $tag              = "";
  my $data             = "";
  my $recordRef        = RepbaseRecord->new();
  my $firstRecordFound = 0;
  while ( <IN> ) {

    ## Ignore spacers ( and RepBase bug where they placed ;'s before tags )
    next if ( /^;?XX/ );

    ## Get file release if it exists
    if ( /^CC\s+RELEASE\s+(\d+)/ ) {
      $this->{'release'} = $1;
    }

    ## Process the end record tag
    if ( /^\/\/\s*$/ || eof( IN ) ) {
      if ( $data ne "" ) {

        # Parse last tag
        _parseTagData( $tag, $data, $recordRef );

        # Save record!
        push @{ $this->{'records'} }, $recordRef;
        $recordRef = RepbaseRecord->new();
      }
      $data = "";
      next;
    }

    ## Capture the tag!
    if ( /^(;?)(\S\S)(?:\s+(\S.*))?/ ) {
      if ( $1 eq ";" ) {

        # Another Repbase error.  Special case to trap it.
        warn
"WARNING: Found error in repbase file.  Invalid tag in line $_.  Recovering...\n";
      }
      $prevTag = $tag;
      $tag     = $2;
      ## Check validity of the tag
      if ( defined $validTagBits{$tag} ) {

        # If this tag group has ended check to see
        # if we have data to parse
        if ( $tag ne $prevTag ) {
          if ( $data ne "" ) {

            # Parse last tag
            _parseTagData( $prevTag, $data, $recordRef );
          }

          # New tag so overwrite old data
          if ( defined $3 ) {
            $data = $3 . "\n";
          }
          else {
            $data = "\n";
          }
        }
        else {

          # This is a tag continuation so
          # append the data
          if ( defined $3 ) {
            $data .= $3 . "\n";
          }
          else {
            $data .= "\n";
          }
        }
      }
      else {
        croak "Error - invalid tag $2 on line:\n$_\n   In file: $filename";
      }
    }
    else {
      if ( /^\s+[\w\s]+\d+$/ ) {
        $data .= $_;
      }
      else {
        print STDERR "Warning...unknown stuff <$_>\n";
      }
    }
  }
  close IN;
}

##-------------------------------------------------------------------------##
## Use: my _parseTagData( $tag, $data, $recRef );
##
##  Parse a Repbase/EMBL data element.  This is a helper function
##  for the general parser.  This function expects data for
##  multiline tags to already be compressed into a single
##  blob.  Ie.
##
##     CC  line1
##     CC  line2
##
##  Should be passed to this function as:
##
##     $tag = "CC"
##     $data = "line1\nline2"
##
##-------------------------------------------------------------------------##
sub _parseTagData {
  my $tag    = shift;
  my $data   = shift;
  my $recRef = shift;
  return if ( $tag ne "ID" && !defined $recRef->getId() );
  if ( $tag eq "ID" ) {

    # IDentification line
    if ( $data =~ /(\S+)\s+(\S+)\;\s+(\S+)\s*\;\s+(\S+)\s*\;\s+(\d+)\s+BP.*/ ) {
      $recRef->setId( $1 );
      $recRef->setDataclass( $2 );
      $recRef->setMolecule( $3 );
      $recRef->pushDivisions( $4 );
    }
    elsif ( $data =~ /(\S+)\s+(\S+)\;\s+(\S+)\s*\;\s+(\S+)\s*\;\s+HUM\s+BP.*/ )
    {
      warn "parseTagData: ID field not to EMBL"
          . " spec \"$_\"...recoverable\".\n";
      $recRef->setId( $1 );
      $recRef->setDataclass( $2 );
      $recRef->setMolecule( $3 );
      $recRef->pushDivisions( $4 );
    }
    else {
      croak "parseTagData: ID field not to EMBL" . " spec \"$data\" from $_\n";
    }
  }
  elsif ( $tag eq "CC" ) {
    if ( $data =~ /RepeatMasker\s+Annotations/i ) {
      if ( $data =~ /Type:[ \t]*(\S+)[ \t]*[\n\r]/i ) {
        $recRef->setRMType( $1 );
      }
      if ( $data =~ /SubType:[ \t]*(\S+)[ \t]*[\n\r]/i ) {
        $recRef->setRMSubType( $1 );
      }

      # TODO: Decide if we want to be consistent with
      #       other lists...ie. place a coma between entities
      if ( $data =~ /Species:[ \t]*([\S \t]+)[ \t]*[\n\r]/i ) {
        my $speciesStr = $1;

        $speciesStr =~ s/\s//g;
        $speciesStr =~ s/(.*),$/$1/g;
        my @species = split( /\,/, $speciesStr );
        foreach my $species ( @species ) {
          if ( $species =~ /Search/ ) { die "Hmmm $speciesStr\n"; }
          $species =~ s/\s//g;
          $recRef->pushRMSpecies( $species );
        }
      }
      if ( $data =~ /Refineable/ ) {
        $recRef->setRMRefineable( 1 );
      }
      if ( $data =~ /SearchStages:\s*([\d\s*\,\s*]+)[\n\r]/i ) {
        my $stageStr = $1;
        $stageStr =~ s/\s//g;
        $stageStr =~ s/(.*),$/$1/g;
        my @stages = split( /\,/, $stageStr );
        foreach my $stage ( @stages ) {
          $recRef->pushRMSearchStages( $stage );
        }
      }
      if ( $data =~ /BufferStages:([^\n\r]*)[\n\r]+/i ) {
        my $bufData = $1;
        while ( $bufData =~ s/\s*(\d+\s*(?:\[\s*\d+\s*-\s*\d+\s*\]\s*)*),?//m )
        {
          my $bufStr = $1;
          $bufStr =~ s/\s//g;
          if ( $bufStr ne "" ) {
            $recRef->pushRMBufferStages( $bufStr );
          }
        }
      }
      if ( $data =~ /Description:\s+(\S.*)/i ) {
        $recRef->setRMDescription( $1 );
      }
    }

    # Coment line
    # Multi-record/multi-line comment allowed
    $recRef->pushComments( $data );
  }
  elsif ( $tag eq "AC" ) {

    # ACession line
    my @accessions = split( /\s*[;\n\r]\s*/, $data );
    foreach my $accession ( @accessions ) {
      $recRef->pushAccessions( $accession );
    }
  }
  elsif ( $tag eq "DT" ) {

    # DaTe line
    # Parse lines
    while ( $data =~ /^(.*)$/gm ) {
      $_ = $1;
      if ( /(\d+)[-\.](\S+)(-\d+)?\s+\(Rel.\s+(\d+(\.\d+)*), Created\)/ ) {
        $recRef->setDateCreated( $_ );
      }
      elsif (
/(\d+)[-\.](\w+)(-\d+)?\s+\(Rel. (\d+(\.\d+)*), Last updated, Version (\-?\d+)\)/
          )
      {
        if ( /Version \-1/ ) {
          s/Version \-1/Version 1/g;
        }
        $recRef->setDateUpdated( $_ );
      }
      elsif ( /(\d+)[-\.](\w+)-(\d+)\s+\(Created\)/ ) {

        # Bug in RepBase "05-MAY-2015 (Created)"
        # Try to guess.
        #   Version major: 2015 = 20 ( so year - 1995 )
        #   Version minor: month
        my %months = (
                       "JAN" => 1,
                       "FEB" => 2,
                       "MAR" => 3,
                       "APR" => 4,
                       "MAY" => 5,
                       "JUN" => 6,
                       "JUL" => 7,
                       "AUG" => 8,
                       "SEP" => 9,
                       "OCT" => 10,
                       "NOV" => 11,
                       "DEC" => 12
        );
        my $majorVersion = $3 - 1995;
        my $minorVersion = $months{$2};
        warn "parseTagData: DT field not to EMBL"
            . " spec \"$_\"...fixing by adding \"Rel. $majorVersion.$minorVersion\".\n";
        s/Created/Rel\. $majorVersion\.$minorVersion, Created/;

        $recRef->setDateUpdated( $_ );
      }
      elsif ( /^DT$/ ) {

        # Empty DT tags...another RepBase bug
        warn "WARNING: Missing DT tag data.  Known RepBase bug...continuing\n";
      }
      else {
        croak "parseTagData: DT field not to EMBL" . " spec \"$data\"\n";
      }
    }
  }
  elsif ( $tag eq "DE" ) {

    # Description line
    $data =~ s/[\n\r]//g;
    $recRef->setDescription( $data );
  }
  elsif ( $tag eq "KW" ) {

    # KeyWord line
    my @keywords = split( /\s*[;\n\r]\s*/, $data );

    # Strip last dot ( used as punctuation in RB EMBL format )
    if ( $keywords[ $#keywords ] =~ /(\S+)\./ ) {
      $keywords[ $#keywords ] = $1;
    }
    foreach my $keyword ( @keywords ) {
      $recRef->pushKeywords( $keyword );
    }
  }
  elsif ( $tag eq "OS" ) {
    ## It looks like GIRI has been using a "CC" line following
    ## the "OS" line to record additional information about the
    ## organism name.  Almost looks like they are using it to store
    ## the common name.  There is an EMBL standard for using the "OC"
    ## line and a pair of parentheses to store the common name.
    $data =~ s/[\n\r]//g;
    if ( ( $data =~ s/\(\s*(.*)\s*\)// ) ) {
      $recRef->setCommonSpeciesName( $1 );
    }
    $recRef->setSpeciesName( $data );
  }
  elsif ( $tag eq "OC" ) {
    my @classifications = split( /\s*[;\n\r]\s*/, $data );

    # Strip last dot ( used as punctuation in RB EMBL format )
    if ( $classifications[ $#classifications ] =~ /(\S+)\./ ) {
      $classifications[ $#classifications ] = $1;
    }
    foreach my $classification ( @classifications ) {
      $recRef->pushClassification( $classification );
    }
  }
  elsif ( $tag eq "RN" ) {
    my $newRef = PubRef->new();
    $newRef->setNumber( $data );
    $recRef->pushReferences( $newRef );
  }
  elsif ( $tag eq "RP" ) {
    my $lastRef =
        $recRef->getReferencesElement( $recRef->getReferencesCount() - 1 );
    $lastRef->setPosition( $data );
  }
  elsif ( $tag eq "RC" ) {
    my $lastRef =
        $recRef->getReferencesElement( $recRef->getReferencesCount() - 1 );
    $lastRef->setComment( $data );
  }
  elsif ( $tag eq "RA" ) {
    my $lastRef =
        $recRef->getReferencesElement( $recRef->getReferencesCount() - 1 );
    $lastRef->setAuthors( $data );
  }
  elsif ( $tag eq "RT" ) {
    my $lastRef =
        $recRef->getReferencesElement( $recRef->getReferencesCount() - 1 );
    $lastRef->setTitle( $data );
  }
  elsif ( $tag eq "RL" ) {
    my $lastRef =
        $recRef->getReferencesElement( $recRef->getReferencesCount() - 1 );
    $lastRef->setLocation( $data );
  }
  elsif ( $tag eq "RX" ) {
    my $lastRef =
        $recRef->getReferencesElement( $recRef->getReferencesCount() - 1 );
    $lastRef->setXref( $data );
  }
  elsif ( $tag eq "DR" ) {
    $recRef->setXref( $data );
  }
  elsif ( $tag eq "SQ" ) {
    my $seq = "";
    while ( $data =~ /^(.*)$/gm ) {
      $_ = $1;
      if (
/Sequence\s+(\d+)\s+BP;(?:\s+(\d+)\s+A;\s+(\d+)\s+C;\s+(\d+)\s+G;\s+(\d+)\s+T;\s+(\d+)\s+other;?)?/
          )
      {
        $recRef->setLength( $1 );
        $recRef->setComposition( 'A',     $2 );
        $recRef->setComposition( 'C',     $3 );
        $recRef->setComposition( 'G',     $4 );
        $recRef->setComposition( 'T',     $5 );
        $recRef->setComposition( 'other', $6 );
      }
      elsif ( /^\s+([acgtbdhvrykmswnxACGTBDHVRYKMSWNX\s]+)\d+$/ ) {
        my $seqLine .= $1;
        $seqLine =~ s/\s//g;
        $seq .= $seqLine;
      }
      elsif ( /^\s+con([acgtbdhvrykmswnxACGTBDHVRYKMSWNX\s]+)\d+$/ ) {

        # Special case: Fix a bug in GIRIs pipeline.  They
        # sometimes leave the prefix "con" at the start of their
        # consensus sequences.  Remove, warn, and move on.
        warn "SQ record for "
            . $recRef->getId()
            . " contains 'con' non-sequence prefix!\n";
        my $seqLine .= $1;
        $seqLine =~ s/\s//g;
        $seq .= $seqLine;
      }
      elsif ( /^\s+([acgtbdhvrykmswnxACGTBDHVRYKMSWNX\s\*lqu]+)\d+$/ ) {

        # Special case: Fix a bug in GIRIs pipeline.  They
        # sometimes have placed "*","l","q" or "u" in the sequence.  Remove,
        # warn, and move on.
        warn "SQ record for "
            . $recRef->getId()
            . " contains *|l|q|u. All known cases of extra characters in RepBase records. Removing offending characters and moving on.  Beware.\n";
        my $seqLine .= $1;
        $seqLine =~ s/[\s\*lqu]//g;
        $seq .= $seqLine;

       # TODO: Probably not sane to just assume these were insertions.  It could
       # be that the overwrote some of the consensus.  Is it still the right
       # length?
      }
      else {
        croak "SQ record for " . $recRef->getId() . " is not valid! $_\n";
      }
    }
    $recRef->setSequence( $seq );
  }
  elsif ( $tag eq "FT" ) {
    my $FTLines = $recRef->getFTLines();
    $recRef->setFTLines( $FTLines . $_ );
  }
  elsif ( $tag eq "RG" ) {

    # Do nothing for now
  }
  elsif ( $tag eq "FH" ) {

    # Do nothing for now
  }
  elsif ( $tag eq "NM" ) {

    # Do nothing for now
  }
  else {
    croak "Not sure how to deal with $tag\n";
  }

}

1;