# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-require 'maasha/digest'
require 'maasha/patternmatcher'
+require 'maasha/bits'
+require 'maasha/backtrack'
+require 'maasha/digest'
#require 'maasha/patscan'
# Residue alphabets
# Quality scores bases
SCORE_PHRED = 33
SCORE_ILLUMINA = 64
+SCORE_MIN = 0
+SCORE_MAX = 40
# Error class for all exceptions to do with Seq.
class SeqError < StandardError; end
class Seq
#include Patscan
include PatternMatcher
+ include BackTrack
+ include Digest
attr_accessor :seq_name, :seq, :type, :qual
- # Method that generates all possible oligos of a specifed length and type.
+ # Class method to instantiate a new Sequence object given
+ # a Biopiece record.
+ def self.new_bp(record)
+ seq_name = record[:SEQ_NAME]
+ seq = record[:SEQ]
+ type = record[:SEQ_TYPE]
+ qual = record[:SCORES]
+
+ self.new(seq_name, seq, type, qual)
+ end
+
+ # Class method that generates all possible oligos of a specifed length and type.
def self.generate_oligos(length, type)
raise SeqError, "Cannot generate negative oligo length: #{length}" if length <= 0
# Method that given a Seq entry returns a FASTA entry (a string).
def to_fasta(wrap = nil)
- raise SeqError, "Missing seq_name" if self.seq_name.nil?
- raise SeqError, "Missing seq" if self.seq.nil?
+ raise SeqError, "Missing seq_name" if self.seq_name.nil? or self.seq_name == ''
+ raise SeqError, "Missing seq" if self.seq.nil? or self.seq.empty?
- seq_name = self.seq_name
- seq = self.seq
+ seq_name = self.seq_name.to_s
+ seq = self.seq.to_s
unless wrap.nil?
seq.gsub!(/(.{#{wrap}})/) do |match|
- match << "\n"
+ match << $/
end
seq.chomp!
end
- ">#{seq_name}\n#{seq}\n"
+ ">" + seq_name + $/ + seq + $/
+ end
+
+ # Method that given a Seq entry returns a FASTQ entry (a string).
+ def to_fastq
+ raise SeqError, "Missing seq_name" if self.seq_name.nil?
+ raise SeqError, "Missing seq" if self.seq.nil?
+ raise SeqError, "Missing qual" if self.qual.nil?
+
+ seq_name = self.seq_name.to_s
+ seq = self.seq.to_s
+ qual = self.qual.to_s
+
+ "@" + seq_name + $/ + seq + $/ + "+" + $/ + qual + $/
end
# Method that generates a unique key for a
def reverse_complement
self.reverse
self.complement
+ self
end
alias :revcomp :reverse_complement
# Method to reverse the sequence.
def reverse
self.seq.reverse!
+ self.qual.reverse! if self.qual
+ self
end
# Method that complements sequence including ambiguity codes.
raise SeqError, "Cannot complement 0 length sequence" if self.length == 0
if self.is_dna?
- self.seq.tr!( 'AGCUTRYWSMKHDVBNagcutrywsmkhdvbn', 'TCGAAYRWSKMDHBVNtcgaayrwskmdhbvn' )
+ self.seq.tr!('AGCUTRYWSMKHDVBNagcutrywsmkhdvbn', 'TCGAAYRWSKMDHBVNtcgaayrwskmdhbvn')
elsif self.is_rna?
- self.seq.tr!( 'AGCUTRYWSMKHDVBNagcutrywsmkhdvbn', 'UCGAAYRWSKMDHBVNucgaayrwskmdhbvn' )
+ self.seq.tr!('AGCUTRYWSMKHDVBNagcutrywsmkhdvbn', 'UCGAAYRWSKMDHBVNucgaayrwskmdhbvn')
else
raise SeqError, "Cannot complement sequence type: #{self.type}"
end
end
+ # Method to determine the Hamming Distance between
+ # two Sequence objects (case insensitive).
+ def hamming_distance(seq)
+ self.seq.upcase.hamming_distance(seq.seq.upcase)
+ end
+
# Method that generates a random sequence of a given length and type.
def generate(length, type)
raise SeqError, "Cannot generate sequence length < 1: #{length}" if length <= 0
seq_new
end
+ # Method to shuffle a sequence readomly inline.
+ def shuffle!
+ self.seq = self.seq.split('').shuffle!.join
+ self
+ end
+
# Method that returns a subsequence of from a given start position
# and of a given length.
def subseq(start, length = self.length - start)
self.subseq(start, length)
end
+ def quality_trim_right(min)
+ raise SeqError, "no sequence" if self.seq.nil?
+ raise SeqError, "no quality score" if self.qual.nil?
+ raise SeqError, "minimum value: #{min} out of range #{SCORE_MIN} .. #{SCORE_MAX}" unless (SCORE_MIN .. SCORE_MAX).include? min
+
+ regex_right = Regexp.new("[#{(SCORE_ILLUMINA).chr}-#{(SCORE_ILLUMINA + min).chr}]+$")
+
+ self.qual.match(regex_right) do |m|
+ self.subseq!(0, $`.length) if $`.length > 0
+ end
+
+ self
+ end
+
+ def quality_trim_left(min)
+ raise SeqError, "no sequence" if self.seq.nil?
+ raise SeqError, "no quality score" if self.qual.nil?
+ raise SeqError, "minimum value: #{min} out of range #{SCORE_MIN} .. #{SCORE_MAX}" unless (SCORE_MIN .. SCORE_MAX).include? min
+
+ regex_left = Regexp.new("^[#{(SCORE_ILLUMINA).chr}-#{(SCORE_ILLUMINA + min).chr}]+")
+
+ self.qual.match(regex_left) do |m|
+ self.subseq!(m.to_s.length, self.length - m.to_s.length) if self.length - m.to_s.length > 0
+ end
+
+ self
+ end
+
+ def quality_trim(min)
+ self.quality_trim_right(min)
+ self.quality_trim_left(min)
+ self
+ end
+
# Method that returns the residue compositions of a sequence in
# a hash where the key is the residue and the value is the residue
# count.
def convert_phred2illumina!
self.qual.gsub!(/./) do |score|
score_phred = score.ord - SCORE_PHRED
- raise SeqError, "Bad Phred score: #{score} (#{score_phred})" unless (0 .. 40).include? score_phred
+ raise SeqError, "Bad Phred score: #{score} (#{score_phred})" unless (0 .. 41).include? score_phred
score_illumina = score_phred + SCORE_ILLUMINA
score = score_illumina.chr
end
projects / biopieces.git / blobdiff