miga-base 0.3.1.7 → 0.3.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: b53d716162f9aedbc64f1e54e02ffc293b16a7e7
-  data.tar.gz: a5c46555329c2da1ba1fd165d423e513a27562ef
+  metadata.gz: c395d2565cacafe425c91a277c03cdeaa3ac9ece
+  data.tar.gz: 1978e8a2df4d0646bc884ce9921bf8de2bc13a25
 SHA512:
-  metadata.gz: c94add412b17de6a932ee247e90ef5682afdf5b61cf09a3b6b9baa64d401da09d29915f6b7a6f39a9e8e6e67ba6e7afb5ed2a982e488805e22f16974cedc9ad7
-  data.tar.gz: 2b8e6fcbdc0b4f1b72e43bb02d47e3fb4618773ef4e86c9644b77926429b6f117cc2d965437704c2540fffd26a8576bdfc299fdd927f7fe2c0f2f33a4c961727
+  metadata.gz: fff79ff971c5fc9e0e0684585b9b35abe5377cff0742c772fc1a7c58005b224ea18b36e7fe489b78c6103f6c4b9ffc90be75e7a385d5fce59a917e2d68c3765a
+  data.tar.gz: cda85047eabd8ba1c76bb5fe690247a3e71583e034f3c5dbef9178776d4c5da02bd3075cfc0e3154e1e97aa69da8ddea85a00ef99efa065e867950c10a860a71

data/actions/ncbi_get.rb

@@ -34,6 +34,8 @@ OptionParser.new do |opt|
   opt.on('--no-version-name',
         'Do not add sequence version to the dataset name.',
         'Only affects --complete and --chromosome.'){ |v| o[:add_version]=v }
+  opt.on('--blacklist PATH',
+        'A file with dataset names to blacklist.'){ |v| o[:blacklist] = v }
   opt.on('--dry', 'Do not download or save the datasets.'){ |v| o[:dry] = v }
   opt.on('-q', '--query',
         'Register the datasets as queries, not reference datasets.'
@@ -135,6 +137,12 @@ if o[:scaffold] or o[:contig]
   end
 end
 
+# Discard blacklisted
+unless o[:blacklist].nil?
+  $stderr.puts "Discarding datasets in #{o[:blacklist]}." unless o[:q]
+  File.readlines(o[:blacklist]).map(&:chomp).each{ |i| ds.delete i }
+end
+
 # Download entries
 $stderr.puts "Downloading #{ds.size} #{ds.size>1 ? "entries" : "entry"}." unless o[:q]
 ds.each do |name,body|

data/lib/miga/common.rb

@@ -1,62 +1,24 @@
 # @package MiGA
 # @license Artistic-2.0
 
-require 'miga/version'
 require 'json'
-require 'tempfile'
-require 'zlib'
+require 'miga/version'
+require 'miga/common/base'
+require 'miga/common/path'
+require 'miga/common/format'
 
 ##
 # Generic class used to handle system-wide information and methods, and parent
 # of all other MiGA::* classes.
 class MiGA::MiGA
+
+  include MiGA::Common
+  
+  extend MiGA::Common::Path
+  extend MiGA::Common::Format
   
   ENV['MIGA_HOME'] ||= ENV['HOME']
 
-  ##
-  # Root path to MiGA (as estimated from the location of the current file).
-  def self.root_path
-    File.expand_path('../../..', __FILE__)
-  end
-
-  ##
-  # Should debugging information be reported?
-  @@DEBUG = false
-
-  ##
-  # Should the trace of debugging information be reported?
-  @@DEBUG_TRACE = false
-
-  ##
-  # Turn on debugging.
-  def self.DEBUG_ON() @@DEBUG=true end
-
-  ##
-  # Turn off debugging.
-  def self.DEBUG_OFF() @@DEBUG=false end
-
-  ##
-  # Turn on debug tracing (and debugging).
-  def self.DEBUG_TRACE_ON
-    @@DEBUG_TRACE=true
-    self.DEBUG_ON
-  end
-
-  ##
-  # Turn off debug tracing (but not debugging).
-  def self.DEBUG_TRACE_OFF
-    @@DEBUG_TRACE=false
-  end
-
-  ##
-  # Send debug message.
-  def self.DEBUG(*args)
-    $stderr.puts(*args) if @@DEBUG
-    if @@DEBUG_TRACE
-      $stderr.puts caller.map{ |v| v.gsub(/^/,'     ') }.join("\n")
-    end
-  end
-
   ##
   # Has MiGA been initialized?
   def self.initialized?
@@ -64,121 +26,6 @@ class MiGA::MiGA
       File.exist?(File.expand_path('.miga_daemon.json', ENV['MIGA_HOME']))
   end
 
-  ##
-  # Tabulates an +values+, and Array of Arrays, all with the same number of
-  # entries as +header+. Returns an Array of String, one per line.
-  def self.tabulate(header, values)
-    fields = [header.map(&:to_s)]
-    fields << fields.first.map{ |h| h.gsub(/\S/, '-') }
-    fields += values.map{ |row| row.map{ |cell| cell.nil? ? '?' : cell.to_s } }
-    clen = fields.map{ |row| row.map(&:length) }.transpose.map(&:max)
-    fields.map do |row|
-      (0 .. clen.size-1).map do |col_n|
-        col_n==0 ? row[col_n].rjust(clen[col_n]) : row[col_n].ljust(clen[col_n])
-      end.join('  ')
-    end
-  end
-
-  ##
-  # Cleans a FastA file in place.
-  def self.clean_fasta_file(file)
-    tmp_fh = nil
-    begin
-      if file =~ /\.gz/
-        tmp_path = Tempfile.new('MiGA.gz').tap(&:close).path
-        tmp_fh = Zlib::GzipWriter.open(tmp_path)
-        fh = Zlib::GzipReader.open(file)
-      else
-        tmp_fh = Tempfile.new('MiGA')
-        tmp_path = tmp_fh.path
-        fh = File.open(file, 'r')
-      end
-      buffer = ''
-      fh.each_line do |ln|
-        ln.chomp!
-        if ln =~ /^>\s*(\S+)(.*)/
-          (id, df) = [$1, $2]
-          tmp_fh.print buffer.wrap_width(80)
-          buffer = ''
-          tmp_fh.puts ">#{id.gsub(/[^A-Za-z0-9_\|\.]/, "_")}#{df}"
-        else
-          buffer << ln.gsub(/[^A-Za-z\.\-]/, '')
-        end
-      end
-      tmp_fh.print buffer.wrap_width(80)
-      tmp_fh.close
-      fh.close
-      FileUtils.cp(tmp_path, file)
-    ensure
-      begin
-        tmp_fh.close unless tmp_fh.nil?
-        File.unlink(tmp_path) unless tmp_path.nil?
-      rescue
-      end
-    end
-  end
-
-  ##
-  # Calculates the average and standard deviation of the sequence lengths in
-  # a FastA or FastQ file (supports gzipped files). The +format+ must be a
-  # Symbol, one of +:fasta+ or +:fastq+. Additional estimations can be
-  # controlled via the +opts+ Hash. Supported options include:
-  # - +:n50+: If true, it also returns the N50 and the median (in bp).
-  # - +gc+: If true, it also returns the G+C content (in %).
-  def self.seqs_length(file, format, opts={})
-    fh = (file =~ /\.gz/) ? Zlib::GzipReader.open(file) : File.open(file, 'r')
-    l = []
-    gc = 0
-    i = 0 # <- Zlib::GzipReader doesn't set $.
-    fh.each_line do |ln|
-      i += 1
-      if (format==:fasta and ln =~ /^>/) or (format==:fastq and (i % 4)==1)
-        l << 0
-      elsif format==:fasta or (i % 4)==2
-        l[l.size-1] += ln.chomp.size
-        gc += ln.scan(/[GCgc]/).count if opts[:gc]
-      end
-    end
-    fh.close
-    
-    o = { n: l.size, tot: l.inject(:+) }
-    o[:avg] = o[:tot].to_f/l.size
-    o[:var] = l.map{ |a| a ** 2 }.inject(:+).to_f/l.size - o[:avg]**2
-    o[:sd]  = Math.sqrt o[:var]
-    o[:gc]  = 100.0*gc/o[:tot] if opts[:gc]
-    if opts[:n50]
-      l.sort!
-      thr = o[:tot]/2
-      pos = 0
-      l.each do |a|
-        pos += a
-        o[:n50] = a
-        break if pos >= thr
-      end
-      o[:med] = o[:n].even? ?
-        0.5*l[o[:n]/2-1,2].inject(:+) : l[(o[:n]-1)/2]
-    end
-    o
-  end
-
-  ##
-  # Path to a script to be executed for +task+. Supported +opts+ are:
-  # - +:miga+ Path to the MiGA home to use. If not passed, the home of the
-  #   library is used).
-  # - +:project+ MiGA::Project object to check within plugins. If not passed,
-  #   only core scripts are supported.
-  def self.script_path(task, opts={})
-    opts[:miga] ||= root_path
-    unless opts[:project].nil?
-      opts[:project].plugins.each do |pl|
-        if File.exist? File.expand_path("scripts/#{task}.bash", pl)
-          opts[:miga] = pl
-        end
-      end
-    end
-    File.expand_path("scripts/#{task}.bash", opts[:miga])
-  end
-
   ##
   # Check if the result files exist with +base+ name (String) followed by the
   # +ext+ values (Array of String).
@@ -188,59 +35,6 @@ class MiGA::MiGA
       File.exist?(base + f) or File.exist?("#{base}#{f}.gz")
     end
   end
-end
-
-##
-# MiGA extensions to the File class.
-class File
 
-  ##
-  # Method to transfer a file from +old_name+ to +new_name+, using a +method+
-  # that can be one of :symlink for File#symlink, :hardlink for File#link, or
-  # :copy for FileUtils#cp_r.
-  def self.generic_transfer(old_name, new_name, method)
-    return nil if exist? new_name
-    case method
-    when :symlink
-      File.symlink(old_name, new_name)
-    when :hardlink
-      File.link(old_name, new_name)
-    when :copy
-      FileUtils.cp_r(old_name, new_name)
-    else
-      raise "Unknown transfer method: #{method}."
-    end
-  end
-end
-
-##
-# MiGA extensions to the String class.
-class String
-
-  ##
-  # Replace any character not allowed in a MiGA name for underscore (_). This
-  # results in a MiGA-compliant name EXCEPT for empty strings, that results in
-  # empty strings.
-  def miga_name
-    gsub(/[^A-Za-z0-9_]/, '_')
-  end
-
-  ##
-  # Is the string a MiGA-compliant name?
-  def miga_name?
-    !(self !~ /^[A-Za-z0-9_]+$/)
-  end
-
-  ##
-  # Replace underscores by spaces or dots (depending on context).
-  def unmiga_name
-    gsub(/_(str|sp|subsp|pv)__/,"_\\1._").tr('_', ' ')
-  end
-
-  ##
-  # Wraps the string with fixed Integer +width+.
-  def wrap_width(width)
-    gsub(/([^\n\r]{1,#{width}})/,"\\1\n")
-  end
 end
 

data/lib/miga/common/base.rb

@@ -0,0 +1,49 @@
+
+class MiGA::MiGA
+  
+  # Class-level
+  class << self
+    ##
+    # Turn on debugging.
+    def DEBUG_ON ; @@DEBUG=true end
+
+    ##
+    # Turn off debugging.
+    def DEBUG_OFF ; @@DEBUG=false end
+
+    ##
+    # Turn on debug tracing (and debugging).
+    def DEBUG_TRACE_ON
+      @@DEBUG_TRACE=true
+      DEBUG_ON()
+    end
+
+    ##
+    # Turn off debug tracing (but not debugging).
+    def DEBUG_TRACE_OFF
+      @@DEBUG_TRACE=false
+    end
+
+    ##
+    # Send debug message.
+    def DEBUG(*args)
+      $stderr.puts(*args) if @@DEBUG
+      $stderr.puts(
+            caller.map{ |v| v.gsub(/^/,'     ') }.join("\n") ) if @@DEBUG_TRACE
+    end
+  end
+
+end
+
+module MiGA::Common
+
+  ##
+  # Should debugging information be reported?
+  @@DEBUG = false
+
+  ##
+  # Should the trace of debugging information be reported?
+  @@DEBUG_TRACE = false
+
+end
+

data/lib/miga/common/format.rb

@@ -0,0 +1,135 @@
+
+require 'tempfile'
+require 'zlib'
+
+module MiGA::Common::Format
+
+  ##
+  # Tabulates an +values+, and Array of Arrays, all with the same number of
+  # entries as +header+. Returns an Array of String, one per line.
+  def tabulate(header, values)
+    fields = [header.map(&:to_s)]
+    fields << fields.first.map{ |h| h.gsub(/\S/, '-') }
+    fields += values.map{ |row| row.map{ |cell| cell.nil? ? '?' : cell.to_s } }
+    clen = fields.map{ |row| row.map(&:length) }.transpose.map(&:max)
+    fields.map do |row|
+      (0 .. clen.size-1).map do |col_n|
+        col_n==0 ? row[col_n].rjust(clen[col_n]) : row[col_n].ljust(clen[col_n])
+      end.join('  ')
+    end
+  end
+
+  ##
+  # Cleans a FastA file in place.
+  def clean_fasta_file(file)
+    tmp_fh = nil
+    begin
+      if file =~ /\.gz/
+        tmp_path = Tempfile.new('MiGA.gz').tap(&:close).path
+        tmp_fh = Zlib::GzipWriter.open(tmp_path)
+        fh = Zlib::GzipReader.open(file)
+      else
+        tmp_fh = Tempfile.new('MiGA')
+        tmp_path = tmp_fh.path
+        fh = File.open(file, 'r')
+      end
+      buffer = ''
+      fh.each_line do |ln|
+        ln.chomp!
+        if ln =~ /^>\s*(\S+)(.*)/
+          (id, df) = [$1, $2]
+          tmp_fh.print buffer.wrap_width(80)
+          buffer = ''
+          tmp_fh.puts ">#{id.gsub(/[^A-Za-z0-9_\|\.]/, "_")}#{df}"
+        else
+          buffer << ln.gsub(/[^A-Za-z\.\-]/, '')
+        end
+      end
+      tmp_fh.print buffer.wrap_width(80)
+      tmp_fh.close
+      fh.close
+      FileUtils.cp(tmp_path, file)
+    ensure
+      begin
+        tmp_fh.close unless tmp_fh.nil?
+        File.unlink(tmp_path) unless tmp_path.nil?
+      rescue
+      end
+    end
+  end
+
+  ##
+  # Calculates the average and standard deviation of the sequence lengths in
+  # a FastA or FastQ file (supports gzipped files). The +format+ must be a
+  # Symbol, one of +:fasta+ or +:fastq+. Additional estimations can be
+  # controlled via the +opts+ Hash. Supported options include:
+  # - +:n50+: If true, it also returns the N50 and the median (in bp).
+  # - +gc+: If true, it also returns the G+C content (in %).
+  def seqs_length(file, format, opts={})
+    fh = (file =~ /\.gz/) ? Zlib::GzipReader.open(file) : File.open(file, 'r')
+    l = []
+    gc = 0
+    i = 0 # <- Zlib::GzipReader doesn't set $.
+    fh.each_line do |ln|
+      i += 1
+      if (format==:fasta and ln =~ /^>/) or (format==:fastq and (i % 4)==1)
+        l << 0
+      elsif format==:fasta or (i % 4)==2
+        l[l.size-1] += ln.chomp.size
+        gc += ln.scan(/[GCgc]/).count if opts[:gc]
+      end
+    end
+    fh.close
+    
+    o = { n: l.size, tot: l.inject(:+) }
+    o[:avg] = o[:tot].to_f/l.size
+    o[:var] = l.map{ |a| a ** 2 }.inject(:+).to_f/l.size - o[:avg]**2
+    o[:sd]  = Math.sqrt o[:var]
+    o[:gc]  = 100.0*gc/o[:tot] if opts[:gc]
+    if opts[:n50]
+      l.sort!
+      thr = o[:tot]/2
+      pos = 0
+      l.each do |a|
+        pos += a
+        o[:n50] = a
+        break if pos >= thr
+      end
+      o[:med] = o[:n].even? ?
+        0.5*l[o[:n]/2-1,2].inject(:+) : l[(o[:n]-1)/2]
+    end
+    o
+  end
+end
+
+##
+# MiGA extensions to the String class.
+class String
+
+  ##
+  # Replace any character not allowed in a MiGA name for underscore (_). This
+  # results in a MiGA-compliant name EXCEPT for empty strings, that results in
+  # empty strings.
+  def miga_name
+    gsub(/[^A-Za-z0-9_]/, '_')
+  end
+
+  ##
+  # Is the string a MiGA-compliant name?
+  def miga_name?
+    !(self !~ /^[A-Za-z0-9_]+$/)
+  end
+
+  ##
+  # Replace underscores by spaces or dots (depending on context).
+  def unmiga_name
+    gsub(/_(str|sp|subsp|pv)__/,"_\\1._").tr('_', ' ')
+  end
+
+  ##
+  # Wraps the string with fixed Integer +width+.
+  def wrap_width(width)
+    gsub(/([^\n\r]{1,#{width}})/,"\\1\n")
+  end
+end
+