# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-# Run Uclust on sequences in the stream.
+# Usearch sequences in the stream against a specified database.
# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
require 'maasha/biopieces'
require 'maasha/fasta'
-
-SORT_LIMIT = 2_000_000_000 # use mergesort for files biggern than 2Gb.
-
-class Uclust
- include Enumerable
-
- def initialize(infile, outfile, options)
- @infile = infile
- @outfile = outfile
- @options = options
- @command = []
- end
-
- # Method that calls Usearch sorting for sorting a FASTA file
- # according to decending sequence length.
- def sort
- # usearch -sort seqs.fasta -output seqs.sorted.fasta
- if File.size(@infile) < SORT_LIMIT
- @command << "usearch --sort #{@infile} --output #{@infile}.sort"
- else
- @command << "usearch --mergesort #{@infile} --output #{@infile}.sort"
- end
-
- execute
-
- File.rename "#{@infile}.sort", @infile
- end
-
- # Method to execute clustering de novo.
- def cluster
- @command << "usearch --cluster #{@infile} --uc #{@outfile} --id #{@options[:identity]}"
-
- execute
- end
-
- # Method to execute database search.
- def usearch
- # usearch --query query.fasta --db db.fasta --uc results.uc --id 0.90 [--evalue E]
- @command << "usearch --query #{@infile} --db #{@options[:database]} --uc #{@outfile} --id #{@options[:identity]}"
- @command << "--evalue #{@options[:e_val]}" if @options.has_key? :e_val
-
- execute
- end
-
- # Method to execute clustering to database plus de novo if not matched.
- def usearch_uclust
- # usearch --cluster seqs_sorted.fasta --db db.fasta --uc results.uc --id 0.90
- @command << "usearch --cluster #{@infile} --db #{@options[:database]} --uc #{@outfile} --id #{@options[:identity]}"
-
- execute
- end
-
- # Method to parse a Uclust .uc file and for each line of data
- # yield a Biopiece record.
- def each
- record = {}
-
- File.open(@outfile, mode="r") do |ios|
- ios.each_line do |line|
- if line !~ /^#/
- fields = line.chomp.split("\t")
-
- record[:REC_TYPE] = "UCLUST"
- record[:TYPE] = fields[0]
- record[:CLUSTER] = fields[1].to_i
- record[:SEQ_LEN] = fields[2].to_i
- record[:IDENT] = fields[3].to_f
- record[:STRAND] = fields[4]
- record[:Q_BEG] = fields[5].to_i
- record[:S_BEG] = fields[6].to_i
- record[:S_END] = fields[6].to_i + fields[2].to_i
- record[:CIGAR] = fields[7]
- record[:Q_ID] = fields[8]
- record[:S_ID] = fields[9]
-
- yield record
- end
- end
- end
-
- self # conventionally
- end
-
- private
-
- # Method to execute a command using a system() call.
- # The command is composed of bits from the @command variable.
- def execute
- @command.unshift "nice -n 19"
- @command << "--rev" if @options[:comp]
- @command << "> /dev/null 2>&1" unless @options[:verbose]
- command = @command.join(" ")
- system(command)
- raise "Command failed: #{command}" unless $?.success?
-
- @command = []
- end
-end
-
-ok_methods = "uclust,usearch,usearch_uclust"
+require 'maasha/usearch'
casts = []
-casts << {:long=>'no_sort', :short=>'n', :type=>'flag', :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'method', :short=>'m', :type=>'string', :mandatory=>true, :default=>"uclust", :allowed=>ok_methods, :disallowed=>nil}
-casts << {:long=>'database', :short=>'d', :type=>'file!', :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'comp', :short=>'c', :type=>'flag', :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'identity', :short=>'i', :type=>'float', :mandatory=>true, :default=>0.9, :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'e_val', :short=>'e', :type=>'float', :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
+casts << {long: 'program', short: 'p', type: 'string', mandatory: false, default: 'global', allowed: "local,global", disallowed: nil}
+casts << {long: 'database', short: 'd', type: 'file!', mandatory: true, default: nil, allowed: nil, disallowed: nil}
+casts << {long: 'identity', short: 'i', type: 'float', mandatory: false, default: nil, allowed: nil, disallowed: nil}
+casts << {long: 'e_val', short: 'e', type: 'float', mandatory: false, default: nil, allowed: nil, disallowed: nil}
+casts << {long: 'cpus', short: 'c', type: 'uint', mandatory: false, default: 1, allowed: nil, disallowed: "0"}
+casts << {long: 'maxaccepts', short: 'm', type: 'uint', mandatory: false, default: 0, allowed: nil, disallowed: nil}
options = Biopieces.options_parse(ARGV, casts)
-# At high identities, around 96% and above, compressed indexes are often more sensitive, faster
-# and use less RAM. Compressed indexes are disabled by default, so I generally recommend that
-# you specify the --slots and --w options when clustering at high identities.
+if options[:program] == 'global'
+ raise ArgumentError, "--identity be specified for global search" unless options[:identity]
+ raise ArgumentError, "--e_val is invalid for global search" if options[:e_val]
+else
+ raise ArgumentError, "--identity or --e_val must be specified" unless options[:identity] or options[:e_val]
+end
+
tmpdir = Biopieces.mktmpdir
infile = File.join(tmpdir, "in.fna")
input.each_record do |record|
output.puts record
- if record.has_key? :SEQ_NAME and record.has_key? :SEQ
+ if record[:SEQ_NAME] and record[:SEQ]
fasta_io.puts Seq.new_bp(record).to_fasta
end
end
end
- uclust = Uclust.new(infile, outfile, options)
- uclust.sort unless options[:no_sort] or options[:method] == "usearch"
+ us = Usearch.new(infile, outfile, options)
- case options[:method].to_s
- when "uclust" then uclust.cluster
- when "usearch" then uclust.usearch
- when "usearch_uclust" then uclust.usearch_uclust
+ case options[:program]
+ when "local" then us.usearch_local
+ when "global" then us.usearch_global
+ else raise "no such program #{options[:program]}"
end
- uclust.each do |record|
+ us.each_hit do |record|
output.puts record
end
end