bio-ngs 0.3.2.alpha.01

data/lib/tasks/history.thor

@@ -0,0 +1,51 @@
+#
+#  
+#
+# Copyright:: Copyright (C) 2011
+#     Francesco Strozzi <francesco.strozzi@gmail.com>
+# License:: The Ruby License
+#
+#
+
+class History < Thor
+  
+
+  
+  history = Bio::Ngs::Record.new(Bio::Ngs::HISTORY_FILE)
+  
+  history.load.each_with_index do |task,index|
+    description = ""
+    if task[:args].is_a?(Array)
+       task[:args].each do |a|
+         if a.is_a?(Hash)
+           a.each_pair do |key,value|
+            description << "#{key} => #{value} "
+           end
+         else
+           description << a+" "
+         end
+       end   
+    else
+      description = task[:args]
+    end
+       
+    # creating History tasks on the fly
+    History.class_eval do
+      desc (index+1).to_s,"Task #{task[:name]} PARAMETERS: #{description}"
+      define_method (index+1).to_s.to_sym do
+        if task[:args].size > 1
+          invoke task[:name], [task[:args][0]], task[:args][1]
+        else
+          invoke task[:name], task[:args]
+        end
+      end
+    end
+  end
+  
+  desc "clear","Wipe out the tasks history"
+  def clear
+   history = Bio::Ngs::Record.new(Bio::Ngs::HISTORY_FILE)
+   history.clear
+  end
+
+end

data/lib/tasks/homology.thor

@@ -0,0 +1,121 @@
+class Homology < Thor
+    
+  class Run < Homology
+    desc "blastn", "Run BlastN"
+    Bio::Ngs::Blast::BlastN.new.thor_task(self,:blastn) do |wrapper, task|
+      wrapper.params = task.options
+      puts wrapper.run :arguments => [file]
+    end
+    
+    desc "blastx", "Run BlastX"
+    Bio::Ngs::Blast::BlastX.new.thor_task(self,:blastx) do |wrapper, task|
+      wrapper.params = task.options
+      puts wrapper.run :arguments => [file]
+    end
+    
+  end
+  
+
+  class Db < Homology
+  
+    desc "init", "Initialize Homology DB"
+    def init
+      if Dir.exists? "db" and Dir.exists? "conf"
+        db = Bio::Ngs::Db.new :homology
+        db.create_tables
+      else
+        puts "No db or conf directory found! Please run 'biongs project:update:annotation'"
+        exit
+      end
+    end
+    
+    desc "export [TABLE]","Export the data from a table to a tab-separated file"
+    method_option :fileout, :type => :string, :desc => "file used to save the output", :required => true
+    def export(table)
+      if Dir.exists? "db"
+        db = Bio::Ngs::Db.new :homology
+        db.export(table,options[:fileout])
+      else
+        puts "No conf directory found! Can't load database connection information"
+        exit
+      end  
+    end
+    
+  end
+
+
+  class Load < Homology
+    
+    desc "blast [FILE]","Parse Blast XML output and load the results into Homology DB"
+    def blast(file)      
+      Bio::Ngs::Homology.blast_import file
+      puts "Parising completed. All the data are now stored into the db.\n"
+    end
+    
+    desc "goa","Import GO Annotation file"
+    method_option :file, :type => :string, :default => "data/goa_uniprot"
+    def goa
+      Bio::Ngs::Homology.goa_import options[:file]
+      puts "Import completed.\n"
+    end
+    
+  end
+  
+  class Report < Homology
+    
+    desc "blast","Output a graphical report on the Blast homology search"
+    method_option :file, :type => :string, :desc => "Read the results from a file and not from the db"
+    method_option :fileout, :type => :string, :desc => "File to write the SVG", :default => "blast_report.svg"
+    def blast
+      db = Bio::Ngs::Db.new :homology
+      evalues = []
+      positive_70 = 0
+      total = BlastOutput.count(:all)
+      positive_70 = BlastOutput.count(:conditions => "positive >= 70")
+      evalue_5 = BlastOutput.count(:conditions => "evalue <= 1e-5")
+      BlastOutput.find(:all).each do |result|
+        evalues << result.evalue
+      end
+      Bio::Ngs::Graphics.bar_charts(["Total mapped","Positive (>=70)","E-value (<=1-e5)"],[total,positive_70,evalue_5],options[:fileout])
+    end
+    
+  end
+  
+  class Convert < Homology
+    
+    
+    desc "blast2text [XML FILE]","Convert Blast output to tab-separated file"
+    method_option :file_out, :type => :string, :required => true, :desc => "File name for report"
+    def blast(file)
+      Bio::Ngs::Homology.blast2text(file,options[:file_out])
+    end
+    
+    desc "go2json", "Convert the GO annotations from the db into a JSON file"
+    method_option :file_out, :type => :string, :default => "data/go_annotations.json", :desc => "File name for JSON"
+    def go2json
+      Bio::Ngs::Homology.go_annotation_to_json(options[:file_out],options[:library])
+    end
+    
+  end
+  
+  class Download < Homology
+
+    desc "uniprot","Download the Uniprot-SwissProt file from UniprotKB"
+    def uniprot
+      Bio::Ngs::Utils.download_and_uncompress("ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/complete/uniprot_sprot.fasta.gz","data/uniprot_sprot.fasta.gz")
+    end
+    
+    desc "goannotation","Download the Uniprot GeneOntology Annotation file"
+    def goannotation
+      Bio::Ngs::Utils.download_and_uncompress("http://cvsweb.geneontology.org/cgi-bin/cvsweb.cgi/go/gene-associations/gene_association.goa_uniprot_noiea.gz?rev=HEAD","data/goa_uniprot.gz")
+    end
+    
+    desc "all", "Download the Uniprot and GO Annotation file"
+    def all
+      invoke :uniprot
+      invoke :goannotation
+    end
+    
+  end
+
+end

data/lib/tasks/ontology.thor

@@ -0,0 +1,93 @@
+class Ontology < Thor
+  
+
+  class Db < Ontology
+  
+    desc "init", "Initialize Ontology DB"
+    def init
+      if Dir.exists? "db" and Dir.exists? "conf"
+        db = Bio::Ngs::Db.new :ontology
+        db.create_tables
+        invoke "ontology:download:go"
+        invoke "ontology:load:go" ["data/gene_ontology.1_2.obo"]
+      else
+        puts "No db or conf directory found! Please run 'biongs project:update:annotation'"
+        exit
+      end
+    end
+    
+    desc "export [TABLE]","Export the data from a table to a tab-separated file"
+    method_option :fileout, :type => :string, :desc => "file used to save the output", :required => true
+    def export(table)
+      if Dir.exists? "db"
+        db = Bio::Ngs::Db.new :ontology
+        db.export(table,options[:fileout])
+      else
+        puts "No conf directory found! Can't load database connection information"
+        exit
+      end
+    end
+    
+  end
+
+
+  class Load < Ontology
+    
+    desc "go [FILE]", "Import GO definition file"
+    def go(file)
+      Bio::Ngs::Ontology.go_import file
+      puts "Import completed.\n"
+    end
+    
+    desc "genego [FILE]", "Import Gene-GO file (JSON)"
+    def genego(file)
+      Bio::Ngs::Ontology.load_go_genes file
+      puts "Import completed"
+    end
+    
+    
+  end
+  
+  class Report < Ontology
+    
+    desc "go","Output a graphical report on the GO for the sequences annotated in the db"
+    def go
+      db = Bio::Ngs::Db.new :ontology
+      ontologies = {}
+      Gene.find(:all).each do |gene|
+        gene.go.each do |ontology|
+          ontologies[ontology.namespace] = Hash.new(0) unless ontologies.has_key? ontology.namespace
+          ontologies[ontology.namespace][ontology.name] += 1
+        end
+      end
+      ontologies.each_pair do |namespace,terms|
+         terms = terms.sort {|a,b| b[1] <=> a[1]}
+         terms.flatten!
+         Bio::Ngs::Graphics.bubble_chart(namespace+"_go.svg",Hash[*terms[0..39]])
+      end
+    end
+    
+  end
+  
+  class Download < Ontology
+    
+    desc "go", "Download the GeneOntology file"
+    def go
+      Bio::Ngs::Utils.download_with_progress(:url => "http://www.geneontology.org/ontology/obo_format_1_2/gene_ontology.1_2.obo", :filename => "data/gene_ontology.1_2.obo")
+    end
+    
+    desc "goslim", "Download the Uniprot GeneOntology Slim file"
+    def goslim
+      Bio::Ngs::Utils.download_with_progress(:url => "http://www.geneontology.org/GO_slims/goslim_goa.obo", :filename => "data/goslim_goa.obo")
+    end
+    
+    desc "all", "Download the GO files"
+    def all
+      invoke :goslim
+      invoke :go
+    end
+
+    
+  end
+
+end

data/lib/tasks/project.thor

@@ -0,0 +1,51 @@
+#
+#  
+#
+# Copyright:: Copyright (C) 2011
+#     Francesco Strozzi <francesco.strozzi@gmail.com>
+# License:: The Ruby License
+#
+#
+
+class Project < Thor
+  include Thor::Actions
+
+  def self.source_root
+     File.expand_path(File.dirname(__FILE__))
+  end
+    
+  attr_accessor :name
+  
+  desc "new [NAME]","Create a new NGS project directory"
+  method_option :type, :type => :string, :desc => "The project type (e.g. annotation)"
+  def new(name)
+    empty_directory name
+    empty_directory File.join("#{name}","data")
+    empty_directory File.join("#{name}","tasks")
+    empty_directory File.join("#{name}","scripts")
+    self.name = name # for template to take the correct values
+    template(File.join("..","templates","README.tt"), "#{name}/README.txt")    
+    
+    if options[:type] == "annotation"
+        invoke "project:update:annotation", [],{:dir => name}
+    else    
+      empty_directory File.join("#{name}","log")
+      empty_directory File.join("#{name}","conf")
+    end
+  end
+  
+  attr_accessor :type
+
+  desc "update [TYPE]", "Update the working dir to a new type of project"
+  method_option :dir, :type => :string
+  def update(type)
+    self.type = type
+    dir = (options[:dir]) ? options[:dir]+"/" : ""
+    empty_directory "#{dir}log"
+    empty_directory "#{dir}conf"
+    empty_directory "#{dir}db"
+    template(File.join("..","templates","db.tt"), "#{dir}conf/#{type}_db.yml")
+  end
+
+  
+end

data/lib/tasks/quality.thor

@@ -0,0 +1,142 @@
+#
+#  
+#
+# Copyright:: Copyright (C) 2011
+#     Francesco Strozzi <francesco.strozzi@gmail.com>
+# License:: The Ruby License
+#
+#
+
+class Quality < Thor
+
+  desc "reads FASTQ", "perform quality check for NGS reads"
+  method_option :width, :type => :numeric, :default => 500
+  method_option :height, :type => :numeric, :default => 500
+  method_option :fileout, :type => :string, :default => "fastq_report.svg"  
+  def reads(fastq)
+    reads = Bio::Ngs::FastQuality.new(fastq)
+    qual = reads.quality_profile
+    Bio::Ngs::Graphics.draw_area(qual,options[:width],options[:height],options[:fileout],"Nucleotide","Quality Score")
+  end
+
+  desc "trim FASTQ", "trim all the sequences"
+  #TODO: create a wrapper
+  method_option :min_size, :type=>:numeric, :default=>20, :aliases => "-l", :desc=>"Minimum length - sequences shorter than this (after trimming)
+                    will be discarded. Default = 0 = no minimum length."
+  method_option :min_quality, :type=>:numeric, :default=>10, :aliases => "-t", :desc=>"Quality threshold - nucleotides with lower 
+                                      quality will be trimmed (from the end of the sequence)."
+  method_option :output, :type=>:string, :aliases => "-o", :desc => "Output file name"
+  def trim(fastq)
+    output_file = options.output || fastq.gsub(/(.*)_(forward|reverse)(.*)/,'\1_trim_\2\3')
+    if output_file==fastq
+      output_file+="_trim"
+    end
+    raise "Input file #{fastq} dosen't exist." unless File.exists?(fastq)
+    unless File.exists?("#{fastq}.txt") #suppose there is a stat file for the input file
+      invoke :fastq_stats, [fastq]
+    end
+    #TODO check the file in input exists
+    trim = Bio::Ngs::Fastx::Trim.new
+    trim.params={min_size:options.min_size, min_quality:options.min_quality, input:fastq, output:output_file}
+    trim.run
+    invoke :fastq_stats, [output_file]
+  end
+  
+  
+  desc "fastq_stats FASTQ", "Reports quality of FASTQ file"
+  method_option :output, :type=>:string, :aliases =>"-o", :desc => "Output file name. default is input file_name with .txt."
+  def fastq_stats(fastq)
+    output_file = options.output || "#{fastq}.txt"
+    stats = Bio::Ngs::Fastx::FastqStats.new
+    stats.params = {input:fastq, output:output_file}
+    stats.run
+    invoke :boxplot, [output_file]
+    invoke :reads_coverage, [output_file]
+  end
+  
+  desc "boxplot FASTQ_QUALITY_STATS", "plot reads quality as boxplot"
+  method_option :title, :type=>:string, :aliases =>"-t", :desc  => "Title (usually the solexa file name) - will be plotted on the graph."
+  method_option :output, :type=>:string, :aliases =>"-o", :desc => "Output file name. default is input file_name with .txt."
+  def boxplot(fastq_quality_stats)
+    output_file = options.output || "#{fastq_quality_stats}.png"
+    boxplot = Bio::Ngs::Fastx::ReadsBoxPlot.new
+    boxplot.params={input:fastq_quality_stats, output:output_file}
+    boxplot.run
+  end
+  
+  desc "reads_coverage FASTQ_QUALITY_STATS", "plot reads coverage in bases"
+  method_option :title, :type=>:string, :aliases =>"-t", :desc  => "Title (usually the solexa file name) - will be plotted on the graph."
+  method_option :output, :type=>:string, :aliases =>"-o", :desc => "Output file name. default is input file_name with .txt."
+  def reads_coverage(fastq_quality_stats)
+    #TODO: port this script to biongs now is only on my server
+    output_file = options.output || "#{fastq_quality_stats}_coverage.png"
+    coverage = Bio::Ngs::Fastx::ReadsCoverage.new
+    coverage.params={input:fastq_quality_stats, output:output_file}
+    coverage.run
+  end
+  
+
+  desc "illumina_b_profile_raw FASTQ", "perform a profile for reads coming fom Illumina 1.5+ and write the report in a txt file"
+  method_option :read_length, :type => :numeric, :required => true
+  method_option :width, :type => :numeric, :default => 500
+  method_option :height, :type => :numeric, :default => 500
+  method_option :fileout, :type => :string, :default => "fastq_report.txt"  
+  def illumina_b_profile_raw(fastq)
+    reads = Bio::Ngs::FastQuality.new(fastq, :fastq_illumina)
+    profile = Array.new(options.read_length,0) #create a default profile setted to zero.
+    quals = reads.track_b_count
+    quals.b_profile.each do |b_item|
+      b_index = b_item[0]
+      b_count = b_item[1]
+      profile[b_index] = b_count if b_index <= options.read_length
+    end
+    File.open(options.fileout,'w') do |f|
+      f.puts "Total reads: #{quals.n_reads}"
+      profile.each_index do |index|
+        f.puts "#{index},#{profile[index]}"
+      end
+    end#File
+  end
+
+  desc "illumina_b_profile_svg FASTQ", "perform a profile for reads coming fom Illumina 1.5+"
+  method_option :read_length, :type => :numeric, :required => true
+  method_option :width, :type => :numeric, :default => 500
+  method_option :height, :type => :numeric, :default => 500
+  method_option :fileout, :type => :string, :default => "fastq_report.svg"  
+  def illumina_b_profile_svg(fastq)
+    reads = Bio::Ngs::FastQuality.new(fastq, :fastq_illumina)
+    profile = Array.new(options.read_length,0) #create a default profile setted to zero.
+    reads.track_b_count.b_profile.each do |b_item|
+      b_index = b_item[0]
+      b_count = b_item[1]
+      profile[b_index] = b_count if b_index <= options.read_length
+    end
+
+    Bio::Ngs::Graphics.draw_area(profile,options[:width],options[:height],options[:fileout], "B distribution", "Nucleotides", "Counts", n_ticks=30)
+  end  
+
+  desc "scatterplot EXPR1 EXPR2 OUTPUT", "plot quantification values as scatterplot in png format"
+  method_option :title, :type=>:string, :aliases =>"-t", :desc  => "Title plotted on the graph."
+  def scatterplot (expr1, expr2, output)
+                                                                                                  
+    [expr1, expr2].each do |file_name|                                                            #controllo sul file!
+      unless File.exists?(file_name)
+             raise "Input file #{file_name} doesn't exist, please insert a valid file name."
+      end
+    end
+    
+    system "sort #{expr1} > tmp_1"      #con system richiami la shell
+    system "sort #{expr2} > tmp_2"
+    File.open("tmp_gnuplot",'w') do |f|
+      f.puts "set title '#{options.title || "Scatter plot NGS"}'"
+      f.puts "set terminal png"
+      f.puts "set output '#{output}.png'"
+      f.puts "plot '< join tmp_1 tmp_2 | head -n -1' using 6:14"
+    end                                           
+    puts "gnuplot tmp_gnuplot"
+    system "cat tmp_gnuplot"
+    system "rm tmp_1 tmp_2 tmp_gnuplot"    
+  end 
+end
+
+