bio-polymarker 1.3.2

data/bin/vcfToPolyMarker.rb

@@ -0,0 +1,82 @@
+#!/usr/bin/env ruby
+
+require 'optparse'
+
+require 'csv'
+require 'bio'
+require 'bio-samtools-wrapper'
+
+$: << File.expand_path(File.dirname(__FILE__) + '/../lib')
+$: << File.expand_path('.')
+path= File.expand_path(File.dirname(__FILE__) + '/../lib/bio-polymarker.rb')
+require path
+
+options = {}
+options[:arm_selection] = Bio::PolyploidTools::ChromosomeArm.getArmSelection("nrgene");
+
+
+OptionParser.new do |opts|
+	opts.banner = "Usage: vcfToPolyMarker.rb [options]"
+
+	opts.on("-c", "--reference FILE", "File with genome reference to use as database") do |o|
+		options[:path_to_contigs] = o
+	end
+
+	opts.on("-a", "--arm_selection #{Bio::PolyploidTools::ChromosomeArm.getValidFunctions.join('|')}", "Function to decide the chromome arm") do |o|
+		tmp_str = o
+		arr = o.split(",")
+		if arr.size == 2
+			options[:arm_selection] = lambda do |contig_name|
+				separator, field = arr
+				field = field.to_i
+				ret = contig_name.split(separator)[field]
+				return ret
+			end
+		else
+			options[:arm_selection] = Bio::PolyploidTools::ChromosomeArm.getArmSelection(o)
+		end
+	end
+
+
+end.parse!
+
+def parseVCFheader(head_line="")
+	##INFO=<ID=NS,Number=1,Type=Integer,Description="Number of samples with data">
+
+	m=/##INFO=<ID=(.+),Number=(.+),Type=(.+),Description="(.+)">/.match(head_line)
+	{:id=>m[1],:number=>m[2],:type=>m[3],:desc=>m[4]}
+
+end
+
+
+
+
+header_info = Hash.new
+ref=options[:path_to_contigs]
+
+fasta_reference_db = Bio::DB::Fasta::FastaFile.new({:fasta=>ref})
+fasta_reference_db.load_fai_entries 
+
+$stdin.each do |line|  
+
+	h = nil
+	h = parseVCFheader(line) if line.start_with? "##INFO"
+
+	header_info[h[:id]] = h[:desc] if h
+	#puts header_info.inspect
+	next if line.start_with? "##"
+	if line.start_with? "#CHROM"
+		arr = line.split
+		arr = arr.drop(9)
+		arr2 = arr.map { |s| [s.clone().prepend('Cov'), s.clone().prepend('Hap') ]}
+		#header += arr2.join("\t")
+		#puts header
+		next
+	end
+	line.chomp!
+	#puts line
+	snp = Bio::PolyploidTools::SNP.parseVCF( line , chr_arm_parser: options[:arm_selection])
+	#puts snp.inspect
+	snp.setTemplateFromFastaFile(fasta_reference_db, flanking_size: 100)
+	puts [snp.gene, snp.chromosome ,snp.to_polymarker_sequence(100)].join(",")
+end

data/bio-polymarker.gemspec

@@ -0,0 +1,227 @@
+# Generated by juwelier
+# DO NOT EDIT THIS FILE DIRECTLY
+# Instead, edit Juwelier::Tasks in Rakefile, and run 'rake gemspec'
+# -*- encoding: utf-8 -*-
+# stub: bio-polymarker 1.3.2 ruby lib
+
+Gem::Specification.new do |s|
+  s.name = "bio-polymarker".freeze
+  s.version = "1.3.2"
+
+  s.required_rubygems_version = Gem::Requirement.new(">= 0".freeze) if s.respond_to? :required_rubygems_version=
+  s.require_paths = ["lib".freeze]
+  s.authors = ["Rob Ellis".freeze]
+  s.date = "2024-09-25"
+  s.description = "Repository of tools developed at Crop Genetics in JIC to work with polyploid wheat".freeze
+  s.email = "rob.ellis@jic.ac.uk".freeze
+  s.executables = ["bfr.rb".freeze, "blast_triads.rb".freeze, "blast_triads_promoters.rb".freeze, "count_variations.rb".freeze, "filter_blat_by_target_coverage.rb".freeze, "filter_exonerate_by_identity.rb".freeze, "find_best_blat_hit.rb".freeze, "find_best_exonerate.rb".freeze, "get_longest_hsp_blastx_triads.rb".freeze, "hexaploid_primers.rb".freeze, "homokaryot_primers.rb".freeze, "mafft_triads.rb".freeze, "mafft_triads_promoters.rb".freeze, "map_markers_to_contigs.rb".freeze, "marker_to_vcf.rb".freeze, "markers_in_region.rb".freeze, "mask_triads.rb".freeze, "polymarker.rb".freeze, "polymarker_capillary.rb".freeze, "polymarker_deletions.rb".freeze, "snp_position_to_polymarker.rb".freeze, "snps_between_bams.rb".freeze, "tag_stats.rb".freeze, "vcfLineToTable.rb".freeze, "vcfToPolyMarker.rb".freeze]
+  s.extra_rdoc_files = [
+    "README.md"
+  ]
+  s.files = [
+    ".travis.yml",
+    "Gemfile",
+    "README.md",
+    "Rakefile",
+    "SECURITY.md",
+    "VERSION",
+    "bin/bfr.rb",
+    "bin/blast_triads.rb",
+    "bin/blast_triads_promoters.rb",
+    "bin/count_variations.rb",
+    "bin/filter_blat_by_target_coverage.rb",
+    "bin/filter_exonerate_by_identity.rb",
+    "bin/find_best_blat_hit.rb",
+    "bin/find_best_exonerate.rb",
+    "bin/get_longest_hsp_blastx_triads.rb",
+    "bin/hexaploid_primers.rb",
+    "bin/homokaryot_primers.rb",
+    "bin/mafft_triads.rb",
+    "bin/mafft_triads_promoters.rb",
+    "bin/map_markers_to_contigs.rb",
+    "bin/marker_to_vcf.rb",
+    "bin/markers_in_region.rb",
+    "bin/mask_triads.rb",
+    "bin/polymarker.rb",
+    "bin/polymarker_capillary.rb",
+    "bin/polymarker_deletions.rb",
+    "bin/snp_position_to_polymarker.rb",
+    "bin/snps_between_bams.rb",
+    "bin/tag_stats.rb",
+    "bin/vcfLineToTable.rb",
+    "bin/vcfToPolyMarker.rb",
+    "bio-polymarker.gemspec",
+    "conf/defaults.rb",
+    "conf/primer3_config/dangle.dh",
+    "conf/primer3_config/dangle.ds",
+    "conf/primer3_config/interpretations/dangle_i.dh",
+    "conf/primer3_config/interpretations/dangle_i.ds",
+    "conf/primer3_config/interpretations/loops_i.dh",
+    "conf/primer3_config/interpretations/loops_i.ds",
+    "conf/primer3_config/interpretations/stack_i.dh",
+    "conf/primer3_config/interpretations/stack_i.ds",
+    "conf/primer3_config/interpretations/stackmm_i_mm.dh",
+    "conf/primer3_config/interpretations/stackmm_i_mm.ds",
+    "conf/primer3_config/interpretations/tetraloop_i.dh",
+    "conf/primer3_config/interpretations/tetraloop_i.ds",
+    "conf/primer3_config/interpretations/triloop_i.dh",
+    "conf/primer3_config/interpretations/triloop_i.ds",
+    "conf/primer3_config/interpretations/tstack2_i.dh",
+    "conf/primer3_config/interpretations/tstack2_i.ds",
+    "conf/primer3_config/interpretations/tstack_i.dh",
+    "conf/primer3_config/interpretations/tstack_i.ds",
+    "conf/primer3_config/interpretations/tstack_tm_inf_i.dh",
+    "conf/primer3_config/interpretations/tstack_tm_inf_i.ds",
+    "conf/primer3_config/loops.dh",
+    "conf/primer3_config/loops.ds",
+    "conf/primer3_config/stack.dh",
+    "conf/primer3_config/stack.ds",
+    "conf/primer3_config/stackmm.dh",
+    "conf/primer3_config/stackmm.ds",
+    "conf/primer3_config/tetraloop.dh",
+    "conf/primer3_config/tetraloop.ds",
+    "conf/primer3_config/triloop.dh",
+    "conf/primer3_config/triloop.ds",
+    "conf/primer3_config/tstack.dh",
+    "conf/primer3_config/tstack2.dh",
+    "conf/primer3_config/tstack2.ds",
+    "conf/primer3_config/tstack_tm_inf.ds",
+    "lib/bio/BFRTools.rb",
+    "lib/bio/BIOExtensions.rb",
+    "lib/bio/PolyploidTools/ChromosomeArm.rb",
+    "lib/bio/PolyploidTools/ExonContainer.rb",
+    "lib/bio/PolyploidTools/Marker.rb",
+    "lib/bio/PolyploidTools/Mask.rb",
+    "lib/bio/PolyploidTools/NoSNPSequence.rb",
+    "lib/bio/PolyploidTools/PrimerRegion.rb",
+    "lib/bio/PolyploidTools/SNP.rb",
+    "lib/bio/PolyploidTools/SNPMutant.rb",
+    "lib/bio/PolyploidTools/SNPSequence.rb",
+    "lib/bio/db/blast.rb",
+    "lib/bio/db/exonerate.rb",
+    "lib/bio/db/primer3.rb",
+    "lib/bio-polymarker.rb",
+    "test/data/7B_amplicon_test.fa",
+    "test/data/7B_amplicon_test.fa.fai",
+    "test/data/7B_amplicon_test_reference.fa",
+    "test/data/7B_amplicon_test_reference.fa.fai",
+    "test/data/7B_marker_test.txt",
+    "test/data/BS00068396_51.fa",
+    "test/data/BS00068396_51_blast.tab",
+    "test/data/BS00068396_51_contigs.aln",
+    "test/data/BS00068396_51_contigs.dnd",
+    "test/data/BS00068396_51_contigs.fa",
+    "test/data/BS00068396_51_contigs.fa.fai",
+    "test/data/BS00068396_51_contigs.fa.nhr",
+    "test/data/BS00068396_51_contigs.fa.nin",
+    "test/data/BS00068396_51_contigs.fa.nsq",
+    "test/data/BS00068396_51_contigs.nhr",
+    "test/data/BS00068396_51_contigs.nin",
+    "test/data/BS00068396_51_contigs.nsq",
+    "test/data/BS00068396_51_exonerate.tab",
+    "test/data/BS00068396_51_for_polymarker.txt",
+    "test/data/BS00068396_51_genes.txt",
+    "test/data/IWGSC_CSS_1AL_scaff_1455974.fa",
+    "test/data/IWGSC_CSS_1AL_scaff_1455974_aln_contigs.fa",
+    "test/data/IWGSC_CSS_1AL_scaff_1455974_aln_contigs.fa.fai",
+    "test/data/LIB1716.bam",
+    "test/data/LIB1716.bam.bai",
+    "test/data/LIB1719.bam",
+    "test/data/LIB1719.bam.bai",
+    "test/data/LIB1721.bam",
+    "test/data/LIB1721.bam.bai",
+    "test/data/LIB1722.bam",
+    "test/data/LIB1722.bam.bai",
+    "test/data/PST130_7067.csv",
+    "test/data/PST130_7067.fa",
+    "test/data/PST130_7067.fa.fai",
+    "test/data/PST130_7067.fa.ndb",
+    "test/data/PST130_7067.fa.nhr",
+    "test/data/PST130_7067.fa.nin",
+    "test/data/PST130_7067.fa.not",
+    "test/data/PST130_7067.fa.nsq",
+    "test/data/PST130_7067.fa.ntf",
+    "test/data/PST130_7067.fa.nto",
+    "test/data/PST130_reverse_primer.csv",
+    "test/data/S22380157.fa",
+    "test/data/S22380157.fa.fai",
+    "test/data/S22380157.vcf",
+    "test/data/S58861868/LIB1716.bam",
+    "test/data/S58861868/LIB1716.sam",
+    "test/data/S58861868/LIB1719.bam",
+    "test/data/S58861868/LIB1719.sam",
+    "test/data/S58861868/LIB1721.bam",
+    "test/data/S58861868/LIB1721.sam",
+    "test/data/S58861868/LIB1722.bam",
+    "test/data/S58861868/LIB1722.sam",
+    "test/data/S58861868/S58861868.fa",
+    "test/data/S58861868/S58861868.fa.fai",
+    "test/data/S58861868/S58861868.vcf",
+    "test/data/S58861868/header.txt",
+    "test/data/S58861868/merged.bam",
+    "test/data/S58861868/merged_reheader.bam",
+    "test/data/S58861868/merged_reheader.bam.bai",
+    "test/data/Test3Aspecific.csv",
+    "test/data/Test3Aspecific_contigs.fa",
+    "test/data/bfr_out_test.csv",
+    "test/data/chr1A_C1145499T/chr1A_C1145499T.csv",
+    "test/data/chr1A_G540414846C/chr1A_G540414846C.csv",
+    "test/data/chr1A_G540414846C/chr1A_G540414846C.fa",
+    "test/data/chr1A_T517634750C/chr1A_T517634750C.csv",
+    "test/data/chr2D_C112180134A/chr2D_C112180134A.csv",
+    "test/data/chr4D_C14473543T/chr4D_C14473543T.csv",
+    "test/data/chr4D_C14473543T/chr4D_C14473543T.fa",
+    "test/data/headerMergeed.txt",
+    "test/data/headerS2238015",
+    "test/data/mergedLibs.bam",
+    "test/data/mergedLibsReheader.bam",
+    "test/data/mergedLibsSorted.bam",
+    "test/data/mergedLibsSorted.bam.bai",
+    "test/data/patological_cases5D.csv",
+    "test/data/primer_3_input_header_test",
+    "test/data/short_primer_design_test.csv",
+    "test/data/some_tests/some_tests.csv",
+    "test/data/test_from_mutant.csv",
+    "test/data/test_iselect.csv",
+    "test/data/test_iselect_reference.fa",
+    "test/data/test_iselect_reference.fa.fai",
+    "test/data/test_primer3_error.csv",
+    "test/data/test_primer3_error_contigs.fa",
+    "test/test_bfr.rb",
+    "test/test_blast.rb",
+    "test/test_exon_container.rb",
+    "test/test_exonearate.rb",
+    "test/test_integration.rb",
+    "test/test_snp_parsing.rb",
+    "test/test_wrong_selection.sh"
+  ]
+  s.homepage = "https://github.com/cb2e6f/bio-polymarker".freeze
+  s.licenses = ["MIT".freeze]
+  s.rubygems_version = "3.1.4".freeze
+  s.summary = "Tool to work with polyploids, NGS and molecular biology".freeze
+
+  if s.respond_to? :specification_version then
+    s.specification_version = 4
+  end
+
+  if s.respond_to? :add_runtime_dependency then
+    s.add_runtime_dependency(%q<bio>.freeze, [">= 1.5.1"])
+    s.add_runtime_dependency(%q<bio-samtools-wrapper>.freeze, [">= 2.7.0"])
+    s.add_runtime_dependency(%q<descriptive_statistics>.freeze, [">= 0"])
+    s.add_runtime_dependency(%q<systemu>.freeze, [">= 2.5.2"])
+    s.add_development_dependency(%q<shoulda>.freeze, [">= 2.10"])
+    s.add_development_dependency(%q<test-unit>.freeze, [">= 0"])
+    s.add_development_dependency(%q<juwelier>.freeze, [">= 0"])
+    s.add_development_dependency(%q<sorted_set>.freeze, [">= 0"])
+  else
+    s.add_dependency(%q<bio>.freeze, [">= 1.5.1"])
+    s.add_dependency(%q<bio-samtools-wrapper>.freeze, [">= 2.7.0"])
+    s.add_dependency(%q<descriptive_statistics>.freeze, [">= 0"])
+    s.add_dependency(%q<systemu>.freeze, [">= 2.5.2"])
+    s.add_dependency(%q<shoulda>.freeze, [">= 2.10"])
+    s.add_dependency(%q<test-unit>.freeze, [">= 0"])
+    s.add_dependency(%q<juwelier>.freeze, [">= 0"])
+    s.add_development_dependency(%q<sorted_set>.freeze, [">= 0"])
+  end
+end
+

data/conf/defaults.rb

@@ -0,0 +1 @@
+path_to_chromosomes="/Users/ramirezr/Documents/PHD/201305_Databases/iwgcs"

data/conf/primer3_config/dangle.dh

@@ -0,0 +1,128 @@
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+-500
+4700
+-4100
+-3800
+0
+0
+0
+0
+0
+0
+0
+0
+-5900
+-2600
+-3200
+-5200
+0
+0
+0
+0
+0
+0
+0
+0
+-2100
+-200
+-3900
+-4400
+0
+0
+0
+0
+0
+0
+0
+0
+-700
+4400
+-1600
+2900
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+-2900
+-4100
+-4200
+-200
+0
+0
+0
+0
+0
+0
+0
+0
+-3700
+-4000
+-3900
+-4900
+0
+0
+0
+0
+0
+0
+0
+0
+-6300
+-4400
+-5100
+-4000
+0
+0
+0
+0
+0
+0
+0
+0
+200
+600
+-1100
+-6900
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0
+0

data/conf/primer3_config/dangle.ds

@@ -0,0 +1,128 @@
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+-1.1
+14.2
+-13.1
+-12.6
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+-16.5
+-7.4
+-10.4
+-15
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+-3.9
+-0.1
+-11.2
+-13.1
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+-0.8
+14.9
+-3.6
+10.4
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+-7.6
+-13
+-15
+-0.5
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+-10
+-11.9
+-10.9
+-13.8
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+-17.1
+-12.6
+-14
+-10.9
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+2.3
+3.3
+-1.6
+-20
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf
+inf

data/conf/primer3_config/interpretations/dangle_i.dh

@@ -0,0 +1,131 @@
+/* Enthalpy of dangling end */
+/* 3' dangling */
+AA_A 0
+AC_A 0
+AG_A 0
+AT_A 0
+AA_C 0
+AC_C 0
+AG_C 0
+AT_C 0
+AA_G 0
+AC_G 0
+AG_G 0
+AT_G 0
+AA_T 5
+AC_T -47
+AG_T 41
+AT_T 38
+CA_A 0
+CC_A 0
+CG_A 0
+CT_A 0
+CA_C 0
+CC_C 0
+CG_C 0
+CT_C 0
+CA_G 59
+CC_G 26
+CG_G 32
+CT_G 52
+CA_T 0
+CC_T 0
+CG_T 0
+CT_T 0
+GA_A 0
+GC_A 0
+GG_A 0
+GT_A 0
+GA_C 21
+GC_C 2
+GG_C 39
+GT_C 44
+GA_G 0
+GC_G 0
+GG_G 0
+GT_G 0
+GA_T 0
+GC_T 0
+GG_T 0
+GT_T 0
+TA_A 7
+TC_A -44
+TG_A 16
+TT_A -29
+TA_C 0
+TC_C 0
+TG_C 0
+TT_C 0
+TA_G 0
+TC_G 0
+TG_G 0
+TT_G 0
+TA_T 0
+TC_T 0
+TG_T 0
+TT_T 0
+/* 5' dangling */
+A_AA 0
+A_AC 0
+A_AG 0
+A_AT 0
+A_CA 0
+A_CC 0
+A_CG 0
+A_CT 0
+A_GA 0
+A_GC 0
+A_GG 0
+A_GT 0
+A_TA 29
+A_TC 41
+A_TG 42
+A_TT 2
+C_AA 0
+C_AC 0
+C_AG 0
+C_AT 0
+C_CA 0
+C_CC 0
+C_CG 0
+C_CT 0
+C_GA 37
+C_GC 40
+C_GG 39
+C_GT 49
+C_TA 0
+C_TC 0
+C_TG 0
+C_TT 0
+G_AA 0
+G_AC 0
+G_AG 0
+G_AT 0
+G_CA 63
+G_CC 44
+G_CG 51
+G_CT 40
+G_GA 0
+G_GC 0
+G_GG 0
+G_GT 0
+G_TA 0
+G_TC 0
+G_TG 0
+G_TT 0
+T_AA -2
+T_AC -6
+T_AG 11
+T_AT 69
+T_CA 0
+T_CC 0
+T_CG 0
+T_CT 0
+T_GA 0
+T_GC 0
+T_GG 0
+T_GT 0
+T_TA 0
+T_TC 0
+T_TG 0
+T_TT 0