bio-ngs 0.3.2.alpha.01

data/lib/bio/ngs/ext/versions.yaml

@@ -0,0 +1,73 @@
+common:
+    libgtextutils:
+        version:  0.6
+        url:      http://hannonlab.cshl.edu/fastx_toolkit/libgtextutils-0.6.tar.bz2
+        basename: libgtextutils-0.6
+        suffix:   tar.bz2
+        desc:     ""
+        type:     source
+    fastx:
+        version:  0.0.13
+        url:      http://hannonlab.cshl.edu/fastx_toolkit/fastx_toolkit-0.0.13.tar.bz2
+        basename: fastx_toolkit-0.0.13
+        suffix:   tar.bz2
+        desc:     "Fastx-toolkit version 0.0.13 requires libgtextutils-0.6 (available here for download). A recent g++ compiler (tested with GNU G++ 4.1.2 and later). The fasta_clipping_histogram tool requires two perl modules: PerlIO::gzip and GD::Graph::bars. The fastx_barcode_splitter tool requires GNU sed. The fastq_quality_boxplot tool requires gnuplot version 4.2 or newer."
+        type:     source
+linux:
+    cufflinks:
+        version:  1.1.0
+        url:      http://cufflinks.cbcb.umd.edu/downloads/cufflinks-1.1.0.Linux_x86_64.tar.gz
+        basename: cufflinks-1.1.0.Linux_x86_64
+        suffix:   tar.gz
+        desc:     ""
+        type:     binary
+    tophat:
+        version:  1.3.2
+        url:      http://tophat.cbcb.umd.edu/downloads/tophat-1.3.2.Linux_x86_64.tar.gz
+        basename: tophat-1.3.2.Linux_x86_64
+        suffix:   tar.gz
+        desc:     ""
+        type:     binary
+    bowtie:
+        version:  0.12.7
+        url:      http://sourceforge.net/projects/bowtie-bio/files/bowtie/0.12.7/bowtie-0.12.7-linux-x86_64.zip/download
+        basename: bowtie-0.12.7-linux-x86_64
+        suffix:   zip
+        desc:     ""
+        type:     binary
+    # sra:
+    #     version: 
+    #     url:      http://trace.ncbi.nlm.nih.gov/Traces/sra/static/sratoolkit.2.1.0-centos_linux64.tar.gz
+    #     basename: sratoolkit.2.1.0-centos_linux64
+    #     suffix:   tar.gz
+    #     desc:     ""
+    #     type:     binary
+osx:
+    cufflinks:
+        version:  1.1.0
+        url:      http://cufflinks.cbcb.umd.edu/downloads/cufflinks-1.1.0.OSX_x86_64.tar.gz
+        basename: cufflinks-1.1.0.OSX_x86_64
+        suffix:   tar.gz
+        desc:     ""
+        type:     binary
+    tophat:
+        version:  1.3.2
+        url:      http://tophat.cbcb.umd.edu/downloads/tophat-1.3.2.OSX_x86_64.tar.gz
+        basename: tophat-1.3.2.OSX_x86_64
+        suffix:   tar.gz
+        desc:     ""
+        type:     binary        
+    bowtie:
+        version:  0.12.7
+        url:      http://sourceforge.net/projects/bowtie-bio/files/bowtie/0.12.7/bowtie-0.12.7-macos-10.5-x86_64.zip/download
+        basename: bowtie-0.12.7-macos-10.5-x86_64
+        suffix:   zip
+        desc:     ""
+        type:     binary
+    # sra:
+    #     version: 
+    #     url:      http://trace.ncbi.nlm.nih.gov/Traces/sra/static/sratoolkit.2.1.0-mac64.tar.gz
+    #     basename: sratoolkit.2.1.0-mac64
+    #     suffix:   tar.gz
+    #     desc:     ""
+    #     type:     binary

data/lib/bio/ngs/graphics.rb

@@ -0,0 +1,189 @@
+#
+#  
+#
+# Copyright:: Copyright (C) 2011
+#     Francesco Strozzi <francesco.strozzi@gmail.com>
+# License:: The Ruby License
+#
+#
+
+require 'rubyvis'
+
+module Bio
+  module Ngs
+    class Graphics
+      
+        def self.draw_area(data,width,height,out=nil,xlabel,ylabel)
+          point = 0
+          max = data.max + 10
+          data = data.map do |d|
+            point += 1
+            OpenStruct.new({:x=> point, :y=> d})
+          end
+          x = pv.Scale.linear(data, lambda {|d| d.x}).range(0, width)
+          y = pv.Scale.linear(0, max).range(0, height);
+          
+          #The root panel
+          vis = pv.Panel.new() do
+            width width
+            height height
+            bottom 20
+            left 50
+            right 10
+            top 5
+
+          # Y-axis and ticks
+            rule do
+              data y.ticks(n_ticks)
+              bottom(y)
+              stroke_style {|d| d!=0 ? "#eee" : "#000"}
+              label(:anchor=>"left") {
+                puts y.inspect
+                text y.tick_format
+              }
+            end
+
+          # X-axis and ticks.
+            rule do
+              data x.ticks()
+              visible {|d| d!=0}
+              left(x)
+              bottom(-5)
+              height(5)
+              label(:anchor=>'bottom') {
+                text(x.tick_format)
+              }
+            end
+            
+          #/* The area with top line. */
+            area do |a|
+              a.data data
+              a.bottom(1)
+              a.left {|d| x.scale(d.x)}
+              a.height {|d| y.scale(d.y)}
+              a.fill_style("rgb(121,173,210)")
+              a.line(:anchor=>'top') {
+                line_width(3)
+              }
+            end
+          end
+          
+          # panel legend and title
+          panel = vis.add(Rubyvis::Panel).
+            width(width-x_padding).
+            height(height)
+            
+          panel.anchor('top').add(Rubyvis::Label).
+            font("20px sans-serif").
+            text(title_label)
+          
+          panel.anchor('bottom').add(Rubyvis::Label).text(xlabel)
+          panel.anchor('left').add(Rubyvis::Label).
+            text_angle(1.5*Math::PI).
+            text(ylabel)
+          
+          
+          vis.render();
+          
+          if out
+            File.open(out,"w") {|f| f.write(vis.to_svg) }
+          else
+            puts vis.to_svg
+          end
+
+      end
+      
+      def self.bubble_chart(fileout,dataset = {}, panel_w = 600, panel_h = 800)
+        colors=Rubyvis::Colors.category10()
+        c=Rubyvis::Colors.category10().by(lambda {|n| n.parent_node})
+
+        vis = Rubyvis::Panel.new
+        .width(panel_w-10)
+        .height(panel_h-10)
+        .bottom(5)
+        .left(5)
+        .right(5)
+        .top(5)
+
+        root=Rubyvis::Dom::Node.new
+        dataset.each_pair do |name,value|
+          child = Rubyvis::Dom::Node.new(value)
+          child.node_name = name
+          root.append_child(child)
+        end
+        root = root.nodes()
+
+        pack=vis.add(pv.Layout.Pack).
+        nodes(root).
+        size(lambda {|n| n.node_value})
+
+        pack.node.add(Rubyvis::Dot).
+        visible( lambda {|n| n.parent_node}).
+        fill_style(lambda {|n|
+          colors.scale(n.parent_node).
+          brighter((n.node_value) / 5.0)
+          }).
+          stroke_style(c)
+
+          pack.node_label.add(Rubyvis::Label).
+          visible( lambda {|n| n.parent_node}).
+          text(lambda {|n| n.node_name})
+        vis.render()
+        File.open(fileout,"w") {|f| f.write vis.to_svg+"\n"}
+      end
+  
+  
+      def self.bar_charts(labels, data, fileout, width = 500, height = 300)
+        
+        x = pv.Scale.linear(0, data.max).range(0, width)
+        y = pv.Scale.ordinal(pv.range(data.size)).split_banded(0, height, 4/5.0)
+
+        #/* The root panel. */
+        vis = pv.Panel.new()
+            .width(width)
+            .height(height)
+            .bottom(20)
+            .left(100)
+            .right(10)
+            .top(5);
+
+        #/* The bars. */
+        bar = vis.add(pv.Bar)
+            .data(data)
+            .top(lambda {y.scale(self.index)})
+            .height(y.range_band)
+            .left(0)
+            .width(x)
+
+        #/* The value label. */
+        bar.anchor("right").add(pv.Label)
+            .text_style("white")
+            .text(lambda {|d| "%0.1f" % d})
+
+        #/* The variable label. */
+        bar.anchor("left").add(pv.Label)
+            .text_margin(5)
+            .text_align("right")
+            .text(lambda { labels[self.index]});
+
+        #/* X-axis ticks. */
+        vis.add(pv.Rule)
+            .data(x.ticks(5))
+            .left(x)
+            .stroke_style(lambda {|d|  d!=0 ? "rgba(255,255,255,.3)" : "#000"})
+          .add(pv.Rule)
+            .bottom(0)
+            .height(5)
+            .stroke_style("#000")
+          .anchor("bottom").add(pv.Label).text(x.tick_format)
+        
+        # X-axis Labels
+        vis.anchor("top").add(Rubyvis::Label).text("Number of sequences")
+        
+        vis.render();
+        File.open(fileout,"w") {|out| out.write vis.to_svg+"\n"}
+      end
+  
+    end
+  end
+end

data/lib/bio/ngs/homology.rb

@@ -0,0 +1,102 @@
+#
+#
+# Copyright:: Copyright (C) 2011
+#     Francesco Strozzi <francesco.strozzi@gmail.com>
+# License:: The Ruby License
+#
+#
+
+module Bio
+  module Ngs
+    class Homology
+      
+      
+      # Method to import a Blast XML output file into a BlastOuput table created according to ActiveRecord model
+      # Params: XML Blast file, YAML file for db connection, optional ActiveRecord models file
+      def self.blast_import(file,yaml_file=nil)
+        db = Bio::Ngs::Db.new :homology,yaml_file
+        inserts = []
+        Bio::Blast::XmlIterator.new(file).to_enum.each do |iter|
+          iter.each do |hit|
+            identity = 0.0
+            positive = 0.0
+            evalue = []
+            length = 0
+            hit.each do |hsp|
+              identity += hsp.identity.to_f
+              positive += hsp.positive.to_f
+              evalue << hsp.evalue
+              length += hsp.align_len
+            end
+            identity =  (identity / length)*100
+            positive = (positive / length)*100
+            evalue = evalue.inject{ |sum, el| sum + el }.to_f / evalue.size
+            inserts << [iter.query_def,hit.hit_id.split('|')[1],hit.hit_def,evalue,identity,positive]
+            if inserts.size == 1000
+              db.insert_many(:blast_outputs,"INSERT INTO blast_outputs(query_id,target_id,target_description,evalue,identity,positive) VALUES(?,?,?,?,?,?)",inserts)
+              inserts = []
+            end  
+          end
+        end
+        db.insert_many(:blast_outputs,"INSERT INTO blast_outputs(query_id,target_id,target_description,evalue,identity,positive) VALUES(?,?,?,?,?,?)",inserts) if inserts.size > 0
+      end
+      
+      def self.blast2text(file_in,file_out)
+        out = File.open(file_out,"w")
+        out.write("Query ID\tTarget ID\tTarget Description\tE-value\tIdentity\tPositive\n")
+        Bio::Blast::XmlIterator.new(file_in).to_enum.each do |iter|
+          iter.each do |hit|
+            identity = 0.0
+            positive = 0.0
+            evalue = []
+            length = 0
+            hit.each do |hsp|
+              identity += hsp.identity.to_f
+              positive += hsp.positive.to_f
+              evalue << hsp.evalue
+              length += hsp.align_len
+            end
+            identity =  (identity / length)*100
+            positive = (positive / length)*100
+            evalue = evalue.inject{ |sum, el| sum + el }.to_f / evalue.size
+            out.write([iter.query_def,hit.hit_id,hit.hit_def,evalue,identity,positive].join("\t")+"\n")
+          end
+        end
+        out.close 
+      end
+
+      
+      # Method to import a GO Annotation file into GoAnnotation table created according to ActiveRecord model
+      # Params: GOA file, YAML file for db connection (optional)
+      def self.goa_import(file,yaml_file=nil)
+        db = Bio::Ngs::Db.new :homology, yaml_file
+        inserts = []
+        File.open(file).each do |line|
+          next if line.start_with? "!"
+          line.chomp!
+          inserts << line.split("\t")[0..14]
+          if inserts.size == 1000
+            db.insert_many(:go_annotations,"INSERT INTO go_annotations(db,entry_id,symbol,qualifier,go_id,db_ref,evidence,additional_identifier,aspect,name,synonym,molecule_type,taxon_id,date,assigned_by) VALUES(?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",inserts)
+            inserts = []
+          end
+        end
+        db.insert_many(:go_annotations,"INSERT INTO go_annotations(db,entry_id,symbol,qualifier,go_id,db_ref,evidence,additional_identifier,aspect,name,synonym,molecule_type,taxon_id,date,assigned_by) VALUES(?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",inserts) if inserts.size > 0
+      end
+      
+      # Method to export the associations among genes and GO and store them into a JSON file that can be imported into the Ontology db
+      # Params: file to write JSON data
+      def self.go_annotation_to_json(file_out,library=nil,yaml_file=nil)
+        db = Bio::Ngs::Db.new :homology, yaml_file
+        ontologies = []
+        BlastOutput.find(:all).each do |result|
+          ontology = Bio::Ngs::Ontology.new result.query_id
+          ontology.go = result.go_annotations.map {|goa| goa.go_id}
+          ontology.library = library
+          ontologies << ontology
+        end
+        File.open(file_out,"w") {|f| f.write ontologies.to_json}
+      end
+      
+    end
+  end
+end

data/lib/bio/ngs/ontology.rb

@@ -0,0 +1,103 @@
+#
+#
+# Copyright:: Copyright (C) 2011
+#     Francesco Strozzi <francesco.strozzi@gmail.com>
+# License:: The Ruby License
+#
+#
+
+module Bio
+  module Ngs
+    class Ontology
+
+      # Method to import a GO OBO file into Go table created according to ActiveRecord model
+      # Params: GO OBO file, YAML file for db connection
+      def self.go_import(file,yaml_file=nil)
+        db = Bio::Ngs::Db.new :ontology,yaml_file
+        inserts = []
+        file = File.open(file)
+        file.each do |line|
+          if line.start_with? "[Term]"
+            block = file.gets("\n\n")
+            is_a = []
+            data = []
+            block.split("\n").each do |elem|
+              if elem.start_with? "id: "
+                data << elem.gsub("id: ","")
+              elsif elem.start_with? "name: "
+                data << elem.gsub("name: ","")
+              elsif elem.start_with? "is_a"
+                is_a << elem.gsub("is_a: ","").split("!").first
+              elsif elem.start_with? "namespace: "
+                data << elem.gsub("namespace: ","")
+              end
+            end
+            data << is_a.join(" ")
+            inserts << data
+            if inserts.size == 1000
+              db.insert_many(:go,"INSERT INTO go(go_id,name,namespace,is_a) VALUES(?,?,?,?)",inserts)
+              inserts = []
+            end
+          end
+        end
+        db.insert_many(:go,"INSERT INTO go(go_id,name,namespace,is_a) VALUES(?,?,?,?)",inserts) if inserts.size > 0
+      end   
+      
+      # Method to lood the Gene-GO associations from a JSON file into the Ontology db
+      # Params: JSON file name, YAML file for db connection (optional)
+      def self.load_go_genes(file,yaml_file=nil)
+        db = Bio::Ngs::Db.new :ontology, yaml_file
+        list = JSON.load File.read(file)
+        ontologies = Bio::Ngs::OntologyCollection.new
+        list.each_with_index do |gene,index|
+          ontologies << Bio::Ngs::Ontology.new(gene["gene_id"],gene["go"],gene["library"])
+        end
+        ontologies.to_db(yaml_file)
+      end
+      
+      
+      attr_accessor :gene_id, :go, :library
+      # Constructor for Bio::Ngs::Ontology instances
+      def initialize(gene_id,go=[],library=nil)
+        @gene_id = gene_id
+        @go = go
+        @library = library
+      end
+      
+      # Method to store a single Bio::Ngs::Ontology object into the Ontology db
+      def to_db(yaml_file=nil)
+        raise RuntimeError,"You must initialize the Ontolgy db with biongs ontology:db:init" if Go.count == 0
+        db = Bio::Ngs::Db.new :ontology,yaml_file
+        g = Gene.create(:gene_id => @gene_id, :library => @library)
+        Go.where({:go_id => @go}).all.each do |go|
+          g.gene_gos.create(:go_id => go.id)
+        end
+      end
+      
+      
+    end
+    
+    # Class to handle collection of Bio::Ngs::Ontology objects. 
+    # It provides a method to store all the gene-GO associations into the Ontology db
+    class OntologyCollection < Array
+      
+      def to_db(yaml_file=nil)
+        db = Bio::Ngs::Db.new :ontology, yaml_file
+        genes = []
+        ontologies = []
+        go = {}
+        Go.find_by_sql("SELECT id, go_id FROM go").each {|g| go[g.go_id] = g.id}
+        self.each_with_index do |gene,index|
+          raise ArgumentError "OntologyCollection can store only Bio::Ngs::Ontology objects!" if gene.class != Bio::Ngs::Ontology
+          genes << [index+1,gene.gene_id,gene.library]
+          gene.go.each {|o| ontologies << [index+1,go[o]] if go[o]}
+        end
+        db.insert_many(:genes,"INSERT INTO genes(id,gene_id,library) VALUES(?,?,?)",genes)
+        db.insert_many(:gene_gos,"INSERT INTO gene_gos(gene_id,go_id) VALUES(?,?)",ontologies)        
+      end
+      
+    end
+    
+    
+  end
+end