create the new files for a rc release

rc_release_bump.rb

#!/usr/bin/env ruby

require "fileutils"

class Assembly

	def initialize(id,dir)
		@id=id
		@dir=dir
	end

	def get_dir
		@latestDir = Dir[@dir+"/assembly/curated/#{@id}.*"].sort_by{ |f| File.mtime(f)}[-1]
		@version = File.extname(@latestDir).to_s[1..-1].to_i+1
		@oldversion = @version-1
		@inPrimary = Dir.glob([@latestDir + "/#{@id}_#{@oldversion}*primary.fa" , @latestDir + "/#{@id}_#{@oldversion}*primary*.fa" , @latestDir + "/#{@id}.#{@oldversion}*primary.fa"])[0].to_s
		@inAlt = Dir.glob([@latestDir + "/#{@id}_#{@oldversion}*haplotigs.fa",@latestDir + "/#{@id}.#{@oldversion}*haplotigs.fa"])[0].to_s
		raise "primary input file is missing" if not File.exists?(@inPrimary)
	end

	def make_new_dir
		@newDir = @dir + "/assembly/curated/"+@id+"."[email protected]_s
		FileUtils.mkpath(@newDir)
		puts "created #{@newDir}"
		outP=File.basename(@inPrimary)

#		outP.sub!("#{@id}_#{(@oldversion).to_s}","#{@id}_#{@version.to_s}")
		outP.sub!("#{@id}.#{(@oldversion).to_s}","#{@id}.#{@version.to_s}")

		if File.exists?(@inAlt)
			outA=File.basename(@inAlt)

#			outA.sub!("#{@id}_#{(@oldversion).to_s}","#{@id}_#{@version.to_s}")
			outA.sub!("#{@id}.#{(@oldversion).to_s}","#{@id}.#{@version.to_s}")
			@outAlt = @newDir +"/"+ outA
		end

		@outPrimary = @newDir +"/" +outP
	end

	def parse_decon
		files = Dir[@dir+"/assembly/draft/#{@id}.*/*.contamination"]
		@blacklist=[]
		files.each{|f|
			File.open(f).each_line{|l|
				if match=l.match(/REMOVE\s+(\S+)/)
					id = match.captures[0]
					@blacklist.push(id.to_s)
				end
			}
		}
		@blacklist.uniq!
	end

	def strip_fasta
		f = File.new(@newDir+"/"+@id+"_"[email protected]_s+".decon_ids","w")
  		@blacklist.each { |e| f.puts(e) }
  		f.puts "scaffold_MT"
  		f.close

  		if @inPrimary == @outPrimary or @inAlt == @outAlt
  			raise "the target directory is the same"
  		end
		a = "/software/grit/bin/fastaremove -f #{@inPrimary} -r #{f.path} > #{@outPrimary}"
		todo=[a]
		if File.exists?(@inAlt)
			todo.push("/software/grit/bin/fastaremove -f #{@inAlt} -r #{f.path} > #{@outAlt}")
		end
		todo.each{|p|
			if not system(p)
				raise "cannot execute: #{p}"
			end
		}

	end

	def move_files
	 	infiles = Dir[@latestDir + "/*"]
	 	infiles.each{|f|
	 		if File.file?(f)
	 			to = @newDir + "/"+ File.basename(f)
	 			to[@id+"_"+(@version-1).to_s]=@id+"_"[email protected]_s
	 			#puts "cp #{f} => #{to}"
	 			FileUtils.cp(f,to)
			end
	 	}
	end
end

# read tab separated values (and skip the ones with a #):
# ilApoCrat1	/lustre/scratch116/tol/projects/darwin/data/insects/Aporia_crataegi
File.open(ARGV[0]).each_line{|l|
	next if l.include? "#"
	id,path = l.chomp!.split
	puts "processing #{l}"
	a=Assembly.new(id,path)
	a.get_dir
	a.make_new_dir
	a.parse_decon
	a.strip_fasta
}

test.conf

ilApoCrat1	/lustre/scratch116/tol/projects/darwin/data/insects/Aporia_crataegi
iySelTumu1	/lustre/scratch116/tol/projects/darwin/data/insects/Seladonia_tumulorum
iySphMoni1	/lustre/scratch116/tol/projects/darwin/data/insects/Sphecodes_monilicornis
iyTenNoth1	/lustre/scratch116/tol/projects/darwin/data/insects/Tenthredo_notha
drMalSylv7	/lustre/scratch116/tol/projects/darwin/data/dicots/Malus_sylvestris
idBelPand1	/lustre/scratch116/tol/projects/darwin/data/insects/Bellardia_pandia
idGymRotn1	/lustre/scratch116/tol/projects/darwin/data/insects/Gymnosoma_rotundatum
drMalDome5	/lustre/scratch116/tol/projects/darwin/data/dicots/Malus_domestica
drMalDome58	/lustre/scratch116/tol/projects/darwin/data/dicots/Malus_domestica
ilEmmMono1	/lustre/scratch116/tol/projects/darwin/data/insects/Emmelina_monodactyla
iiLimLuna2	/lustre/scratch116/tol/projects/darwin/data/insects/Limnephilus_lunatus