Class: PredictORF

Inherits:
Object
  • Object
show all
Includes:
Bio::Big::FrameCodonHelpers
Defined in:
lib/bigbio/sequence/predictorf.rb

Constant Summary

Constants included from Bio::Big::FrameCodonHelpers

Bio::Big::FrameCodonHelpers::START_CODONS, Bio::Big::FrameCodonHelpers::STOP_CODONS

Instance Method Summary collapse

Constructor Details

#initialize(id, descr, seq, trn_table = nil) ⇒ PredictORF

Returns a new instance of PredictORF.



84
85
86
87
88
89
90
91
92
93
94
95
 
# File 'lib/bigbio/sequence/predictorf.rb', line 84

def initialize id, descr, seq, trn_table = nil
  @id        = id
  @descr     = descr
  @seq       = seq.gsub(/\s/,'')
  @trn_table = trn_table
  @startcodons =  # FIXME: this should be linked properly
    if trn_table == nil or trn_table == 0
      START_CODONS
    else # prokaryote
      ['ATG','TTG','CTG','AUG','UUG','CUG']
    end
end

Instance Method Details

#longest_startstop(minsize = 0) ⇒ Object

Return the longest ORF that has a START codon (see startstop) Returns nil if none is found



136
137
138
 
# File 'lib/bigbio/sequence/predictorf.rb', line 136

def longest_startstop minsize=0
  startstop(minsize).first
end

#startstop(minsize = 30) ⇒ Object

Return a list of predicted ORFs with :minsize AA’s. The ORF’s are between START and STOP codons (ATG, TTG, CTG and AUG, UUG and CUG for now, a later version should use the EMBOSS translation table).



127
128
129
130
131
132
 
# File 'lib/bigbio/sequence/predictorf.rb', line 127

def startstop minsize=30
  stopstop(minsize).find_all { | orf | 
    # p [orf.nt.seq[0..2].upcase,@startcodons.include?(orf.nt.seq[0..2].upcase)]
    @startcodons.include?(orf.nt.seq[0..2].upcase)
  }
end

#stopstop(minsize = 30) ⇒ Object

Return a list of predicted ORFs with :minsize AA’s. The ORF’s are between STOP codons (so sequences without a proper START codon are included)



100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
 
# File 'lib/bigbio/sequence/predictorf.rb', line 100

def stopstop minsize=30
  type = "XX"
  orfs = []
  translate = Nucleotide::Translate.new(@trn_table)
  aa_frames = translate.aa_6_frames(@seq)
  # p @seq
  # pp aa_frames
  num = 0
  aa_frames.each do | aa_frame |
    frame = aa_frame[:frame]
    aa = aa_frame[:sequence]
    aa_start = 0
    aa.split(/\*/).each do | candidate |
      if candidate.size >= minsize and candidate.size > 0
        orf = ORF.new(num,type,@id,@descr,@seq,frame,aa_start*3,candidate)
        orfs.push orf
        num += 1
      end
      aa_start += candidate.size + 1
    end
  end
  orfs
end