Module: Bio::NCBIDB::Common

Defined in:: lib/bio/db/genbank/common.rb

Overview

Description

This module defines a common framework among GenBank, GenPept, RefSeq, and DDBJ. For more details, see the documentations in each genbank/*.rb files.

References

Constant Summary

DELIMITER =

RS = "\n//\n"

TAGSIZE =

Instance Method Summary (collapse)

- (Object) acc_version

Returns the first part of the VERSION record as "ACCESSION.VERSION" String.
- (Object) accession

Returns the ACCESSION part of the acc_version.
- (Object) accessions

ACCESSION -- Returns contents of the ACCESSION record as an Array.
- (Object) comment

COMMENT -- Returns contents of the COMMENT record as a String.
- (Object) common_name (also: #vernacular_name)
- (Object) definition

DEFINITION -- Returns contents of the DEFINITION record as a String.
- (Object) features

FEATURES -- Returns contents of the FEATURES record as an array of Bio::Feature objects.
- (Object) gi

Returns the second part of the VERSION record as a "GI:#######" String.
- (Common) initialize(entry)

A new instance of Common.
- (Object) keywords

KEYWORDS -- Returns contents of the KEYWORDS record as an Array of Strings.
- (Object) locus

LOCUS -- Locus class must be defined in child classes.
- (Object) nid

NID -- Returns contents of the NID record as a String.
- (Object) organism
- (Object) origin

ORIGIN -- Returns contents of the ORIGIN record as a String.
- (Object) references

REFERENCE -- Returns contents of the REFERENCE records as an Array of Bio::Reference objects.
- (Object) segment

SEGMENT -- Returns contents of the SEGMENT record as a "m/n" form String.
- (Object) source

SOURCE -- Returns contents of the SOURCE record as a Hash.
- (Object) taxonomy
- (Object) version

Returns the VERSION part of the acc_version as a Fixnum.
- (Object) versions

VERSION -- Returns contents of the VERSION record as an Array of Strings.

Instance Method Details

- (`Object`) acc_version

Returns the first part of the VERSION record as "ACCESSION.VERSION" String.



57
58
59

# File 'lib/bio/db/genbank/common.rb', line 57

def acc_version
  versions.first.to_s
end

- (`Object`) accession

Returns the ACCESSION part of the acc_version.



62
63
64

# File 'lib/bio/db/genbank/common.rb', line 62

def accession
  acc_version.split(/\./).first.to_s
end

- (`Object`) accessions

ACCESSION -- Returns contents of the ACCESSION record as an Array.



46
47
48

# File 'lib/bio/db/genbank/common.rb', line 46

def accessions
  field_fetch('ACCESSION').strip.split(/\s+/)
end

- (`Object`) comment

COMMENT -- Returns contents of the COMMENT record as a String.

# File 'lib/bio/db/genbank/common.rb', line 199

def comment
  str = get('COMMENT').to_s.sub(/\ACOMMENT     /, '')
  str.gsub!(/^ {12}/, '')
  str.chomp!
  str
end

- (`Object`) common_name Also known as: vernacular_name



120
121
122

# File 'lib/bio/db/genbank/common.rb', line 120

def common_name
  source['common_name']
end

- (`Object`) definition

DEFINITION -- Returns contents of the DEFINITION record as a String.



40
41
42

# File 'lib/bio/db/genbank/common.rb', line 40

def definition
  field_fetch('DEFINITION')
end

- (`Object`) features

FEATURES -- Returns contents of the FEATURES record as an array of Bio::Feature objects.

# File 'lib/bio/db/genbank/common.rb', line 209

def features
  unless @data['FEATURES']
    ary = []
    in_quote = false
    get('FEATURES').each_line do |line|
      next if line =~ /^FEATURES/

      # feature type  (source, CDS, ...)
      head = line[0,20].to_s.strip

      # feature value (position or /qualifier=)
      body = line[20,60].to_s.chomp

      # sub-array [ feature type, position, /q="data", ... ]
      if line =~ /^ {5}\S/
        ary.push([ head, body ])

      # feature qualifier start (/q="data..., /q="data...", /q=data, /q)
      elsif body =~ /^ \// and not in_quote		# gb:IRO125195
        ary.last.push(body)
        
        # flag for open quote (/q="data...)
        if body =~ /="/ and body !~ /"$/
          in_quote = true
        end

      # feature qualifier continued (...data..., ...data...")
      else
        ary.last.last << body

        # flag for closing quote (/q="data... lines  ...")
        if body =~ /"$/
          in_quote = false
        end
      end
    end

    ary.collect! do |subary|
      parse_qualifiers(subary)
    end

    @data['FEATURES'] = ary.extend(Bio::Features::BackwardCompatibility)
  end
  if block_given?
    @data['FEATURES'].each do |f|
      yield f
    end
  else
    @data['FEATURES']
  end
end

- (`Object`) gi

Returns the second part of the VERSION record as a "GI:#######" String.



72
73
74

# File 'lib/bio/db/genbank/common.rb', line 72

def gi
  versions.last
end

- (`Common`) initialize(entry)

A new instance of Common

Returns:

(Common) —

a new instance of Common



30
31
32

# File 'lib/bio/db/genbank/common.rb', line 30

def initialize(entry)
  super(entry, TAGSIZE)
end

- (`Object`) keywords

KEYWORDS -- Returns contents of the KEYWORDS record as an Array of Strings.



84
85
86

# File 'lib/bio/db/genbank/common.rb', line 84

def keywords
  @data['KEYWORDS'] ||= fetch('KEYWORDS').chomp('.').split(/; /)
end

- (`Object`) locus

LOCUS -- Locus class must be defined in child classes.



35
36
37

# File 'lib/bio/db/genbank/common.rb', line 35

def locus
  # must be overrided in each subclass
end

- (`Object`) nid

NID -- Returns contents of the NID record as a String.



78
79
80

# File 'lib/bio/db/genbank/common.rb', line 78

def nid
  field_fetch('NID')
end

- (`Object`) organism



125
126
127

# File 'lib/bio/db/genbank/common.rb', line 125

def organism
  source['organism']
end

- (`Object`) origin

ORIGIN -- Returns contents of the ORIGIN record as a String.

# File 'lib/bio/db/genbank/common.rb', line 263

def origin
  unless @data['ORIGIN']
    ori, seqstr = get('ORIGIN').split("\n", 2)
    seqstr ||= ""
    @data['ORIGIN'] = truncate(tag_cut(ori))
    @data['SEQUENCE'] = seqstr.tr("0-9 \t\n\r\/", '')
  end
  @data['ORIGIN']
end

- (`Object`) references

REFERENCE -- Returns contents of the REFERENCE records as an Array of Bio::Reference objects.

# File 'lib/bio/db/genbank/common.rb', line 136

def references
  unless @data['REFERENCE']
    ary = []
    toptag2array(get('REFERENCE')).each do |ref|
      hash = Hash.new
      subtag2array(ref).each do |field|
        case tag_get(field)
        when /REFERENCE/
          if /(\d+)(\s*\((.+)\))?/m =~ tag_cut(field) then
            hash['embl_gb_record_number'] = $1.to_i
            if $3 and $3 != 'sites' then
              seqpos = $3
              seqpos.sub!(/\A\s*bases\s+/, '')
              seqpos.gsub!(/(\d+)\s+to\s+(\d+)/, "\\1-\\2")
              seqpos.gsub!(/\s*\;\s*/, ', ')
              hash['sequence_position'] = seqpos
            end
          end
        when /AUTHORS/
          authors = truncate(tag_cut(field))
          authors = authors.split(/, /)
          authors[-1] = authors[-1].split(/\s+and\s+/) if authors[-1]
          authors = authors.flatten.map { |a| a.sub(/,/, ', ') }
          hash['authors']	= authors
        when /TITLE/
          hash['title']	= truncate(tag_cut(field))
          # CHECK Actually GenBank is not demanding for dot at the end of TITLE
          #+ '.'
        when /JOURNAL/
          journal = truncate(tag_cut(field))
          if journal =~ /(.*) (\d+) \((\d+)\), (\d+-\d+) \((\d+)\)$/
    	hash['journal']	= $1
    	hash['volume']	= $2
    	hash['issue']	= $3
    	hash['pages']	= $4
    	hash['year']	= $5
          else
    	hash['journal'] = journal
          end
        when /MEDLINE/
          hash['medline']	= truncate(tag_cut(field))
        when /PUBMED/
          hash['pubmed']	= truncate(tag_cut(field))
        when /REMARK/
          hash['comments'] ||= []
          hash['comments'].push truncate(tag_cut(field))
        end
      end
      ary.push(Reference.new(hash))
    end
    @data['REFERENCE'] = ary.extend(Bio::References::BackwardCompatibility)
  end
  if block_given?
    @data['REFERENCE'].each do |r|
      yield r
    end
  else
    @data['REFERENCE']
  end
end

- (`Object`) segment

SEGMENT -- Returns contents of the SEGMENT record as a "m/n" form String.



90
91
92

# File 'lib/bio/db/genbank/common.rb', line 90

def segment
  @data['SEGMENT'] ||= fetch('SEGMENT').scan(/\d+/).join("/")
end

- (`Object`) source

SOURCE -- Returns contents of the SOURCE record as a Hash.

# File 'lib/bio/db/genbank/common.rb', line 96

def source
  unless @data['SOURCE']
    name, org = get('SOURCE').split('ORGANISM')
    org ||= ""
    if org[/\S+;/]
      organism = $`
      taxonomy = $& + $'
    elsif org[/\S+\./]				# rs:NC_001741
      organism = $`
      taxonomy = $& + $'
    else
      organism = org
      taxonomy = ''
    end
    @data['SOURCE'] = {
      'common_name'	=> truncate(tag_cut(name)),
      'organism'	=> truncate(organism),
      'taxonomy'	=> truncate(taxonomy),
    }
    @data['SOURCE'].default = ''
  end
  @data['SOURCE']
end

- (`Object`) taxonomy



129
130
131

# File 'lib/bio/db/genbank/common.rb', line 129

def taxonomy
  source['taxonomy']
end

- (`Object`) version

Returns the VERSION part of the acc_version as a Fixnum



67
68
69

# File 'lib/bio/db/genbank/common.rb', line 67

def version
  acc_version.split(/\./).last.to_i
end

- (`Object`) versions

VERSION -- Returns contents of the VERSION record as an Array of Strings.



52
53
54

# File 'lib/bio/db/genbank/common.rb', line 52

def versions
  @data['VERSION'] ||= fetch('VERSION').split(/\s+/)
end

Module: Bio::NCBIDB::Common

Overview

Description

References

Constant Summary

Instance Method Summary (collapse)

Instance Method Details

- (Object) acc_version

- (Object) accession

- (Object) accessions

- (Object) comment

- (Object) common_name Also known as: vernacular_name

- (Object) definition

- (Object) features

- (Object) gi

- (Common) initialize(entry)

- (Object) keywords

- (Object) locus

- (Object) nid

- (Object) organism

- (Object) origin

- (Object) references

- (Object) segment

- (Object) source

- (Object) taxonomy

- (Object) version

- (Object) versions