seqtrimnext 2.0.29

data/lib/seqtrimnext/classes/graph_stats.rb

@@ -0,0 +1,74 @@
+# $: << '/Users/dariogf/progs/ruby/gems/scbi_plot/lib'
+
+require 'scbi_plot'
+# require 'gnu_plot_graph'
+
+class GraphStats
+
+  def initialize(stats,initial_stats=nil)
+    #load stats
+    init_stats=initial_stats
+    
+    if init_stats.nil?
+      r=File.read(File.join(OUTPUT_PATH,'initial_stats.json'))
+      init_stats= JSON::parse(r)
+    end
+    # puts init_stats.to_json
+    #r=File.read(File.join(File.dirname(__FILE__),'stats.json'))
+    if !File.exists?('graphs')
+      Dir.mkdir('graphs')
+    end
+    @stats=stats
+
+    @stats.each do |plugin_name,plugin_value|
+      # get plugin class
+      begin
+        plugin_class = Object.const_get(plugin_name)
+      rescue Exception => e
+        # puts "RESCUE",e.message,e.backtrace
+        plugin_class = Plugin
+      end
+          
+
+      plugin_value.keys.each do |stats_name|
+        puts "Plotting #{stats_name} from #{plugin_name}"
+        # if graph is not ignored
+        if !plugin_class.graph_ignored?(stats_name)
+
+          x=[]
+          y=[]
+
+          # get filename
+          file_name=File.join('graphs',plugin_class.get_graph_filename(plugin_name,stats_name)+'.png')
+
+          # create new graph object
+          plot=ScbiPlot::Histogram.new(file_name,plugin_class.get_graph_title(plugin_name,stats_name))
+
+          plugin_class.auto_setup(plugin_value[stats_name],stats_name,x,y)
+
+          # puts plugin_class.name.to_s
+          # plot_setup returns true if it has already handled the setup of the plot, if not, handle here
+          if !plugin_class.plot_setup(plugin_value[stats_name],stats_name,x,y,init_stats,plot)
+            if !x.empty? && !y.empty? && (x.length==y.length)
+
+              plot.x_label= "Length"
+              plot.y_label= "Count"
+
+              plot.add_x(x)
+              plot.add_y(y)
+              
+              plot.do_graph
+            end
+            
+          end
+
+          # if !x.empty? && !y.empty? && (x.length==y.length)
+          #   
+          # end
+        end
+      end
+    end
+
+  end
+
+end

data/lib/seqtrimnext/classes/install_database.rb

@@ -0,0 +1,43 @@
+require 'open-uri'
+
+class InstallDatabase
+  
+  
+  def initialize(type,db_path)
+    
+    
+    types=['core','cont_bacteria','cont_fungi','cont_mitochondrias','cont_plastids','cont_ribosome']
+    
+    if types.include?(type)
+      
+      if !File.exists?(db_path)
+        FileUtils.mkdir_p(db_path)
+      end
+      
+      remote_db_url="http://www.scbi.uma.es/downloads/#{type}_db.zip"
+      local_path=File.join(db_path,'core_db.zip')
+      puts "Install databases: #{type}"
+    
+      download_and_unzip(remote_db_url,local_path)
+      
+    else
+      puts "Unknown database #{type}"
+      puts "Available databases:"
+      puts types.join("\n")
+    end
+  end
+  
+  def download_and_unzip(from_url,to_file)
+    puts "Downloading databases from #{from_url} to #{to_file}"
+    
+    open(to_file, "w+") { |f| f.write(open(from_url).read)}
+    
+    puts "Unzipping #{to_file}"
+    
+    # unzip and remove
+    # `cd #{File.dirname(to_file)};unzip #{to_file}; rm #{to_file}`
+    `cd #{File.dirname(to_file)};unzip #{to_file}; rm #{to_file}`
+    
+  end
+  
+end

data/lib/seqtrimnext/classes/install_requirements.rb

@@ -0,0 +1,123 @@
+#########################################
+# Author:: Almudena Bocinos Rioboo
+# This class provided the methods to check if the necesary software is installed in the user system
+ #########################################
+
+class InstallRequirements 
+  
+ 
+  def initialize 
+    @external_requirements = {}
+    @ruby_requirements = {}  
+    load_requirements
+    
+    
+  end
+  
+   def check_install_requirements   
+		  res = true
+		  
+		  errors = check_system_requirements
+		  
+		  if !errors.empty?
+		  
+		    $stderr.puts ' Unable to find these external requeriments:'
+		    errors.each do |error|
+		      $stderr.puts '   -' + error
+		      res = false
+		    end #end each
+		    
+		  end #end if
+
+		  errors = check_ruby_requirements
+		  if !errors.empty?
+		    $stderr.puts ' Unable to find these Ruby requeriments:'
+		    errors.each do |error|
+		      $stderr.puts '   -' + error
+		      res = false
+		    end #end each
+		  end #end if
+		  
+      return res
+  end     
+
+
+  
+private
+  
+  def check_system_requirements
+           
+      errors=[]
+        @external_requirements.each do |cmd,msg|
+          if !system("which #{cmd} > /dev/null ")
+            errors.push "It's necessary to install #{cmd}. " + msg
+          end
+        end
+      
+     return errors
+  end
+  
+  def check_ruby_requirements(install=true)
+      errors=[]
+
+      
+      
+        @ruby_requirements.each do |cmd,msg|
+          if !system("gem list #{cmd} | grep #{cmd} > /dev/null")
+            if install
+              puts "Are you sure you wan't to install #{cmd} gem? ([Y/n]):"
+              res=stdin.readline
+              if res.chomp.upcase!='N'
+                system("echo gem install #{cmd}")
+              end
+            else
+              errors.push "It's necessary to install #{cmd}. Issue a: gem install #{cmd} " + msg
+            end
+          end
+        end
+      
+      return errors
+  end
+   
+   
+   
+  # seqtrim's requirements are specified here
+  def load_requirements
+   
+    @external_requirements['blastn']= "You need to install Blast+ 2.2.24 or greater and make sure it is available in your path (export PATH=$PATH:path_to_blast).\nYou can download it from ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/LATEST/"
+    @external_requirements['cd-hit-454']= "Download from http://code.google.com/p/cdhit/downloads/list"
+    @external_requirements['gnuplot']= "Download from http://www.gnuplot.info/download.html"
+    
+    # @external_requirements['pepe']= ""
+    
+                
+    # @ruby_requirements = { 'n2array' => "" ,    
+    @ruby_requirements['narray'] = ''
+    @ruby_requirements['gnuplot'] = ''
+    @ruby_requirements['term-ansicolor'] = ''
+    @ruby_requirements['xml-simple'] = ''
+    @ruby_requirements['scbi_blast'] = ''
+    @ruby_requirements['scbi_drb'] = ''
+    @ruby_requirements['scbi_fasta'] = ''
+    @ruby_requirements['scbi_fastq'] = ''
+    @ruby_requirements['scbi_plot'] = ''
+    @ruby_requirements['scbi_math'] = ''
+    # @ruby_requirements['scbi_fastq2'] = ''
+    
+  end # end  def
+  
+  def install
+    
+    # gem install gnuplot
+    # gem install narray
+    # gem install scbi_blast
+    # gem install scbi_drb
+    # gem install scbi_fasta
+    # gem install scbi_fastq
+    # gem install term-ansicolor
+    # gem install xml-simple
+    
+    
+  end
+  
+end

data/lib/seqtrimnext/classes/list_db.rb

@@ -0,0 +1,49 @@
+
+#List all entries in a DB, by name
+#
+#list all DB names if db is ALL
+
+class ListDb
+
+def initialize(path,db)
+    
+		filename=File.join(path,'formatted',db)
+	  if File.exists?(filename)
+	  
+				f = File.open(filename)
+		
+				f.grep(/^>(.*)$/) do |line|
+					puts $1
+				end
+				f.close
+		else
+				puts "File #{filename} doesn't exists"
+				puts ''
+				puts "Available databases:"
+				puts '-'*20
+				d=Dir.glob(File.join(path,'formatted','*.fasta'))
+   	  	d.entries.map{|e| puts File.basename(e)}
+
+				
+#		cmd= "grep '^>' #{File.join(path,'formatted',db+'.fasta')}"
+		
+#		system(cmd)
+	  end
+
+end
+
+def self.list_databases(path)
+  res = []
+  
+  if File.exists?(path)
+    d=Dir.glob(File.join(path,'formatted','*.fasta'))
+  
+	  res = d.entries.map{|e| File.basename(e)}
+  end
+  return res
+	
+	
+end
+
+
+end

data/lib/seqtrimnext/classes/make_blast_db.rb

@@ -0,0 +1,113 @@
+
+class MakeBlastDb
+  
+def initialize(dir)
+
+  @db_folder = dir
+  @status_folder = File.join(@db_folder,'status_info')
+  @formatted_folder = File.join(@db_folder,'formatted')
+  
+  update_dbs
+end
+
+def catFasta(path_start,path_end)
+  $LOG.debug("Cat of #{path_start}")
+  
+  # system("cat #{path_start} > #{path_end}")
+  system("cat /dev/null > #{path_end}")
+       
+  system("for i in `find #{path_start} -type f ! -name '.*'`; do echo cat of $i; cat $i >> #{path_end}; echo \"\n\" >> #{path_end}; done")
+  
+end
+
+def dirEmpty?(path_db)
+
+  folder2=Dir.open("#{path_db}")
+  
+  ignore = ['.','..','.DS_Store']
+  
+  res = folder2.entries - ignore
+  
+  return res.empty?  
+end
+
+  def merge_db_files(path_db, db_name, formatted_folder)
+		if !dirEmpty?(path_db)
+		  #hay que hacer el cat solo cuando cambian los ficheros que hay en subfolder1
+		  formatted_file = File.join(formatted_folder, db_name+'.fasta')
+		  catFasta(File.join(path_db),formatted_file)
+		end
+  end
+
+def self.format_db(path_db, db_name, formatted_folder)
+
+    #hay que hacer el cat solo cuando cambian los ficheros que hay en subfolder1
+    formatted_file = File.join(formatted_folder, db_name+'.fasta')
+		cmd = "makeblastdb -in #{formatted_file} -parse_seqids -dbtype nucl >> logs/formatdb.log"
+		system(cmd)
+		$LOG.info(cmd)
+
+end
+
+#---------------------------------------------------------------------------------------------------
+# Check if files for DataBase have been updated, and only when that has happened, makeblastdb will run   
+# Consideres the next directories structure:     
+# 
+#     @dir is the main directory                               
+#     @dir/folder0  is the directoy where will be storaged the DB created/updated
+#     @dir/folder0/subfolder1 is where are storaged all the fasta files of the type subfolder1 
+#     @dir/update is where register the log for each subfolder1, to check if DB has been updated
+#---------------------------------------------------------------------------------------------------
+def update_dbs
+
+  FileUtils.mkdir_p(@status_folder)
+  FileUtils.mkdir_p(@formatted_folder)
+     
+  ignore_folders=['.','..','status_info','formatted']
+
+	$LOG.info("Checking Blast databases at #{@db_folder} for updates")
+ 
+  dbs_folder=Dir.open(@db_folder)
+  
+  #if all file_update.entries is in folder1.entries then cat db/* > DB , make blast, guardar ls nuevo 
+  dbs_folder.entries.each do |db_name|
+  	
+    db_folder=File.join(@db_folder,db_name)
+		if (!ignore_folders.include?(db_name) and File.directory?(db_folder))
+ 			
+ 			#puts "Checking #{db_name} in #{db_folder}"
+ 	
+     #path_db = File.join(@dir,db_folder)
+			
+			# set status files    
+    	new_status_file = File.join(@status_folder,'new_'+db_name+'.txt')
+    	old_status_file = File.join(@status_folder,'old_'+db_name+'.txt')
+    
+      cmd = "ls -lR #{db_folder} > #{new_status_file}"
+      $LOG.debug(cmd)
+      # list new status tu new_status_file
+      # system("ls -lR #{File.join(db_folder,'*')} > #{new_status_file}")
+      system(cmd)
+      
+      # if new and old statuses files changed, then reformat
+      if (!(File.exists?(old_status_file)) || !system("diff -q #{new_status_file} #{old_status_file} > /dev/null ") || !File.exists?(File.join(@formatted_folder,db_name+'.fasta')))
+  					
+						$LOG.info("Database #{db_name} modified. Merging and formatting")
+						
+  					merge_db_files(db_folder,db_name,@formatted_folder)
+  										    
+            MakeBlastDb.format_db(db_folder,db_name,@formatted_folder)
+            
+            # rename new_status_file to replace the old one
+            system("mv #{new_status_file} #{old_status_file}")
+       end
+    	
+    end
+    
+  end #end folder1.entries
+ 
+end
+
+end
+
+

data/lib/seqtrimnext/classes/one_blast.rb

@@ -0,0 +1,41 @@
+#!/usr/bin/env ruby
+
+class OneBlast
+
+def initialize(database, blast_type = 'blastp')
+
+		@blast_type = blast_type
+    @database = database
+    @c=0 
+end
+
+
+def do_blast(seq_fasta)
+
+	@f = File.new('one_blast_aux.fasta','w+')
+	@f.puts ">SEQNAME_"+@c.to_s
+	@f.puts seq_fasta
+	@c = @c+1			
+	@f.close
+
+    cmd = '~blast/programs/x86_64/bin/blastall -p '+@blast_type+' -d '+@database + ' -i one_blast_aux.fasta -o one_blast_aux.out'
+        #puts cmd
+        system(cmd)
+
+    res =''
+    File.open('one_blast_aux.out').each_line { |line|
+
+        res = line
+	
+
+    }
+
+end
+
+def close
+
+end
+
+end
+
+