vardetect-vc 0.0.4 → 0.0.5

data/bin/vardetect_vc

@@ -0,0 +1,46 @@
+#!/usr/bin/env ruby
+require 'rubygems'
+require 'vardetect-vc'
+require "vardetect-vc/lib"
+
+
+sample_id = 'G3418'
+working = 'vcqa'
+
+path = '/Users/soup/Desktop/chula/data'
+ref_file = File.join(path,'ref','hg19.fa')
+
+
+snp_array_file = File.join(path,working,"#{sample_id}_gt_report.txt")
+sam_tool_file = File.join(path,working,"#{sample_id}_SNP_Indel_ANNO.csv")
+sam_file = File.join(path,working,'bam',"#{sample_id}.remdup.uniqMap.TS.bam")
+vc_raw_file = File.join(path,working,'snv-vc',"#{sample_id}_vc_raw_snv.tsv")
+
+
+st_file = File.join(path,working,sample_id,'st_snv.tsv')
+vc_file = File.join(path,working,sample_id,'vc_snv.csv')
+snp_file = File.join(path,working,sample_id,'snpar_snv.csv')
+
+st_ft_file = File.join(path,working,sample_id,'st_filtered_snv.csv')
+vc_ft_file = File.join(path,working,sample_id,'vc_filtered_snv.csv')
+snp_ft_file = File.join(path,working,sample_id,'snpar_filtered_snv.csv')
+
+
+com_vc_file = File.join(path,working,sample_id,'vc_compared.csv')
+com_st_file = File.join(path,working,sample_id,'st_compared.csv')
+com_all_file = File.join(path,working,sample_id,'all_compared.csv')
+
+combine_all_file = File.join(path,working,sample_id,'combine_all.csv')
+analysed_file = File.join(path,working,sample_id,'summary.csv')
+
+# 1. call vc
+params = "snv -ref #{ref_file} -sam #{sam_file} -vc_out #{vc_raw_file} -core 8"
+# params = "snv-range -ref #{ref_file} -sam #{sam_file} -vc_out #{vc_raw_file} -core 8 -chr 1 -start 168665645 -stop 170706648 -debug true"
+params = params.split
+# 156553003
+# 
+# 
+# 170706648
+
+Vardetect::Vc.hi
+Vardetect::Vc.exec params

data/bin/vardetect_vc_batch

@@ -0,0 +1,9 @@
+#!/usr/bin/env ruby
+require 'rubygems'
+require 'vardetect-vc'
+require "vardetect-vc/lib"
+
+
+
+Vardetect::Vc.hi
+Vardetect::Vc.exec ARGV

data/lib/vardetect-vc/call_combine.rb

@@ -0,0 +1,367 @@
+module Vardetect
+module Vc
+
+
+def self.call_combine_all params
+  # ====================================================================================================
+  # 4. All comparison
+  # ====================================================================================================
+  
+   map_vc = []
+   map_st = {}
+   map_snp = {}
+   
+
+   vc_file =params['vc_out']
+   st_file =params['st_out']
+   snp_file =params['snp_out']
+   
+   com_out =params['com_out']
+   
+   f_st = File.open(st_file,'r')
+   f_vc = File.open(vc_file,'r')
+   f_snp = File.open(snp_file,'r')
+   
+   com_output = File.open(com_out,'w')
+   
+   puts "Indexing SNP Array SNV"
+   while str = f_snp.gets
+     v = str.strip.split(',')
+     map_snp[v[0]] = v
+   end
+   
+   puts map_snp.keys[0..10]
+   
+   puts "Indexing SamTools SNV"
+      while str = f_st.gets
+        v = str.strip.split(',')
+        map_st[v[0]] = v
+      end
+   
+    puts map_st.keys[0..10]
+      
+      puts "Indexing VC SNV"
+         while str = f_vc.gets
+           v = str.split(',')
+      
+         
+      
+           # if (v[2]==v[3] and alt.to_f/dep > 0.1) or v[2]!=v[3]   
+           #              map_vc<<v
+           #            end
+           #            
+           # if (v[2]==v[3] and alt.to_f/dep > 0.1) or v[2]!=v[3]   
+              map_vc<<v
+            # end
+           
+           
+         end
+         
+         puts map_vc.size
+         
+         # puts map_vc[0..10]
+      
+        
+        
+        
+        
+        # map_vc.sort!{|a,b| a[1].to_i<=>b[1].to_i }
+            
+            all=0
+            vc_count=0  
+            st_count=0
+            vc_het_rate=0
+            st_het_rate=0
+            final = []
+            
+            
+            for i in  map_vc
+               pos = i[0]
+               # puts pos
+               if snp = map_snp[pos]
+                 
+            
+                 vc_alt = i[5].to_i
+                 vc_dep = i[4].to_i+vc_alt
+                 
+                 vc_alt_a = '-'
+                 vc_het = 0
+                 
+                     
+                 snp_rs = snp[1]
+                 snp_a1 = snp[2]
+                 snp_a2 = snp[3]
+                 
+                 snp_a = "#{snp_a1}/#{snp_a2}"
+                 snp_het = 0 
+                 snp_het = 1 if snp_a1!=snp_a2 
+                 snp_gc = snp[4]
+                 snp_clust = snp[5]
+                 snp_cmp = 1
+                 
+                 snp_check = false
+                 if snp_a.strip !='-/-' 
+                 all+=1 
+                 snp_check = true
+                 end
+                 st_check = false
+                 
+                 vc_het_t = 'hom'
+                 snp_het_t = 'hom'
+                 snp_het_t = 'het' if snp_a1!=snp_a2 
+                 
+                 vc_het_ratio = vc_alt.to_f/vc_dep
+                 if vc_het_ratio >= 0.15
+                   vc_alt_a = i[3] 
+                   vc_het = 1
+                   vc_het_t = 'het'
+                 end
+                      
+                 str = "#{i[0]},#{i[1]},#{i[2]},#{i[2]}/#{vc_alt_a},#{vc_het_t},#{vc_dep},#{vc_alt},#{vc_het_ratio},"
+                 str += "#{snp_rs},#{snp_a},#{snp_het_t},#{snp_gc},#{snp_clust},"
+                 
+                 
+                 
+                 
+                 if st = map_st[pos]
+                   st_het = 0 
+                   st_het = 1 if st[5] =='het'
+                   str += st[4..-1].join(",")
+                    if st_het==snp_het and snp_check
+                    st_het_rate+=1
+                    st_check =true
+                    end
+                    st_count+=1
+                 end
+                 vc_check = false
+                 vc_check = true if vc_het==snp_het and snp_check
+               
+                 if snp_check and vc_check
+                   vc_het_rate+=1 
+                 end 
+                 vc_count+=1
+                
+                 cmp = [snp_check,vc_check,st_check,str]
+                 
+                 # final << cmp
+                 com_output.puts cmp.join(',')
+                 
+                 
+               end
+            end
+            
+        
+        puts "All #{all} is not -/-"
+        
+        puts "VC Hetero Rate : #{vc_het_rate}"
+        
+        puts "ST Hetero Rate : #{st_het_rate}"
+   
+   f_st.close
+   f_vc.close
+   f_snp.close
+   com_output.close
+   
+  
+end
+
+
+def self.analyse_combine params
+  
+
+  # index columns
+  idx = {}
+  %w{snp vc st pos ref vc_a1 vc_a vc_het vc_dep vc_alt vc_r snp_ann snp_a snp_het snp_gc snp_clus st_ref st_a2 st_qua st_dep st_alt st_pos}.each_with_index do |i,index|
+    idx[i.to_sym] = index
+  end
+
+  # constant
+  snp_filter = 0.75
+  alt_ratio = 0.1
+
+
+  gcid = idx[:snp_gc]
+  hid = idx[:snp_het] 
+  depid= idx[:vc_dep]
+
+
+  combine_file =params['com_out']
+  analyse_file =params['out']
+  f_list = File.open(combine_file,'r')
+  out  = File.open(analyse_file,'w')
+
+  all_list = []
+  list = []
+
+  het_list = []
+  hom_list = []
+
+
+  # filter all by gc score snp array 
+  while str = f_list.gets
+    # puts str
+    v = str.strip.split(',')
+    all_list<<v
+    if v[gcid].to_f>snp_filter     
+      list<<v
+    end
+  end
+
+  # filter set of snp array 's het hom
+  for i in list
+    if i[hid]=='het'  
+      het_list<<i
+    else
+      hom_list<<i
+    end
+  end
+
+
+
+  out.puts "SNP Array overlap data "
+  out.puts "Filter GC score at,#{snp_filter}"
+  out.puts "Total snv,#{all_list.size}"
+  out.puts "Total snv not pass filter,#{all_list.size-list.size}"
+  out.puts "Total hetero snv,#{het_list.size}"
+  out.puts "Total homo snv,#{hom_list.size}"
+
+
+
+  vc_het_het = []
+  vc_het_hom = []
+  vc_hom_hom = []
+  vc_hom_het = []
+
+  vchid = getidx :vc_het
+  vcrid = getidx :vc_r
+  [vchid]=='het' or
+  for i in list
+
+    if i[hid]=='het'
+      if  i[vcrid].to_f >= alt_ratio
+        vc_het_het<<i
+      else
+        vc_het_hom<<i
+      end  
+    else
+      if  i[vcrid].to_f >= alt_ratio
+        vc_hom_het<<i
+      else
+        vc_hom_hom<<i
+      end  
+    end
+  end
+
+
+  # matrix=[[vc_het_het.size,vc_het_hom.size],[vc_hom_hom.size,vc_hom_het.size]]
+
+
+
+  require 'erb'
+  outtmp = File.open(analyse_file+".html",'w')
+
+  template = ERB.new(File.open('index.html.erb','r').read)
+
+
+  outtmp.puts template.result(binding)
+
+  outtmp.close
+
+  `open #{analyse_file+".html"}`
+
+
+
+  out.close
+  f_list.close
+end
+
+
+def self.getidx id
+  
+  idx = {}
+  %w{snp vc st pos ref vc_a1 vc_a vc_het vc_dep vc_alt vc_r snp_ann snp_a snp_het snp_gc snp_clus st_ref st_a2 st_qua st_dep st_alt st_pos}.each_with_index do |i,index|
+    idx[i.to_sym] = index
+  end
+  return idx[id]
+  
+end
+
+
+def self.sampling list
+  
+  gcid = getidx :snp_gc
+  depid = getidx :vc_dep
+  vcrid = getidx :vc_r
+
+  
+  # =====================================================================
+  # sampling gc score value from all data set
+     gc_bin = []
+     gc_range = 100
+     gc_max = 1
+     for i in list
+       j = i[gcid].to_f
+       k = (j * gc_range / gc_max).to_i 
+       gc_bin[k]=0 unless gc_bin[k]
+       gc_bin[k]+=1 
+     end
+     
+   # =====================================================================
+   # sampling dep score value from all data set
+      d_bin = []
+      d_range = 100
+      d_max = 500
+    
+      for i in list
+        j = i[depid].to_i
+        k = (j * d_range / d_max).to_i 
+        d_bin[k]=0 unless d_bin[k]
+        d_bin[k]+=1 
+      end
+
+      # =====================================================================
+      # sampling gc score value from all data set
+        
+      alt_bin = []
+      alt_range = 100
+      alt_max = 0.5
+      dep_range = 100
+      dep_max = 500
+      
+    
+      for i in list
+        
+        j = i[vcrid].to_f
+        d = i[depid].to_f
+  
+        k = (j * alt_range / alt_max).to_i 
+        alt_bin[k]=0 unless alt_bin[k]
+        alt_bin[k]+=1 
+  
+      end
+      
+      
+      dot = []
+      
+      for i in list
+        
+        j = i[vcrid].to_f
+        d = i[depid].to_f
+        k = (j * alt_range / alt_max).to_i 
+        k2 = (d * dep_range / dep_max).to_i 
+        
+        dot << "[#{d},#{j}]"
+        
+        
+      end
+      
+      
+      
+      
+      return {:gc=>gc_bin,:dep=>d_bin,:alt=>alt_bin,:dot=>dot}
+     
+end
+
+
+end
+end

data/lib/vardetect-vc/call_prepare.rb

@@ -0,0 +1,122 @@
+
+module Vardetect
+module Vc
+  
+
+  def self.call_prepare_st params
+    # ====================================================================================================
+    # 2. ST preparing filtered chromosome
+    # ====================================================================================================
+    count = 0
+    start = false
+    sam_tool_file =params['st']
+    snp_file =params['st_out']
+
+
+    st_output = File.open(snp_file,'w')
+    CSV.foreach(sam_tool_file) do |row|
+     if start 
+
+        i = row[0][3..-1]
+        id = i.to_i
+        id = i if id == 0
+        key = "#{id}-#{row[1]}"
+        st_output.puts key+","+row.join(",")
+
+       count +=1
+     else
+       start=true
+     end
+    end
+    st_output.close
+
+  end
+
+  def self.call_prepare_snp params
+    # ====================================================================================================
+    # 3. SNP preparing filtered chromosome
+    # ====================================================================================================
+
+    snp_array_file =params['snp']
+    snp_file =params['snp_out']
+
+    count = 0
+    list = []
+    start = false
+    run = false
+
+    file = File.open(snp_array_file,'r')
+    output = File.open snp_file,'w'
+    map = {}
+    while str = file.gets #and count<100
+      str.strip!
+      if str=='[Data]' and !start
+        start = true
+      elsif start
+        s = str.split
+        if s[0] =='Sample' and !run
+          run = true
+        else
+
+            key = "#{s[1]}-#{s[2]}"
+             map[key] = s
+
+        end
+
+      end
+      count+=1
+    end
+      # 
+      map.keys.sort.each do |k|
+          v = map[k]
+          output.puts "#{k},"+v[3..-1].join(",")
+      end
+
+    output.close   
+
+
+
+  end
+
+  def self.call_prepare_vc params
+
+    vc_raw_file =params['vc_raw']
+    vc_file =params['vc_out']
+
+    count = 0
+    list = []
+
+    file = File.open(vc_raw_file,'r')
+    output = File.open vc_file,'w'
+    map = {}
+    while str = file.gets #and count<100
+      str.strip!
+      s = str.split
+      # puts str
+      begin
+      if str
+
+          alt = s[6].to_i  
+          dep = s[5].to_i + alt
+          if ( alt.to_f/dep > 0.1) or s[2]!=s[3] 
+            i = s[0][3..-1]
+            key = "#{i}-#{s[1]}"
+              map[key] = s
+          end
+        end
+      rescue
+        puts str
+      end
+    end
+      # 
+      map.keys.sort.each do |k|
+          v = map[k]
+          output.puts "#{k},"+v[2..-1].join(",")
+      end
+    output.close 
+  end
+
+
+
+end
+end

data/lib/vardetect-vc/call_snv.rb

@@ -0,0 +1,493 @@
+module Vardetect
+module Vc
+
+
+
+def self.form seq, cigar
+  ix = cigar.split(/S|M|I|D/) 
+  index = 0
+  id = 0 
+  str = ''
+  st = 0
+  indel = []
+  # puts seq
+  for i in ix
+    index+=i.size
+    cmd = cigar[index]
+    # puts "#{cmd} #{i.to_i}"
+    index+=1
+    i = i.to_i
+    case cmd
+    when 'S'
+      st = i 
+      id += i 
+    when 'I'
+      indel<<{:pos=>id-st,:type=>'ins',:size=>i,:cigar=>cigar}
+      id+=i
+    when 'D'
+      str+='N'*i
+      indel<<{:pos=>id-st,:type=>'del',:size=>i,:cigar=>cigar}
+    when 'M'
+      str += seq[id..id+i-1] if seq[id..id+i-1]
+      id+=i
+    end
+    # puts "Ref:#{seq}"
+    # puts "Seq:#{str}"
+    # puts "ID #{id}"
+    # puts
+  end
+  return str,indel
+end
+
+def self.call_sam_snv sam,chr,chr_start,chr_stop, debug=nil
+  
+  unless debug
+    show = false
+    indel_show = false
+    var_show = false
+  else
+      s = debug.split('|')
+      show = s.index('show')
+      indel_show =  s.index('indel')
+      var_show =  s.index('var')
+  end
+ 
+  
+  l = nil
+  snps = []
+  indels = []
+  indels_map = {}
+
+  puts "#{chr} #{chr_start} - #{chr_stop}"
+  seq = sam.fetch_reference(chr,chr_start,chr_stop+1000)   
+  sam.load_index
+  
+  upseq = seq.upcase
+      
+  
+  
+  als = []
+
+  
+  als = sam.fetch(chr, chr_start, chr_stop)  
+  
+  profile = []
+  list = []
+  start = nil
+  alleles = {'A'=>0,'C'=>1,'G'=>2,'T'=>3}
+  inv_alleles = {0=>'A',1=>'C',2=>'G',3=>'T'}
+  
+         
+  als.each{|i|
+    
+    # puts "#{i.pos}"
+           
+               unless start
+               start = i.pos  
+               end
+             
+               if i.pos!=start
+               
+                 while start<i.pos 
+                   del = []
+                 
+                   s = ''
+                   ref = seq[start-chr_start]
+                   print "#{start}\t#{ref}\t#{list.size}\t" if show
+                   ref = ref.upcase
+                 
+                 
+                  p = [0,0,0,0]
+                  q = [0,0,0,0]
+                  
+                   list.each_with_index do|l,index|
+                
+                   
+                     c = l.seq[start-l.pos]
+                     cq = l.qual[start-l.pos]
+                           
+                    # a = seq[start-1..start-1-5]
+                    # b =seq[start-l.pos..start-l.pos-5]
+                    # puts "#{a} - #{b}"
+                  
+                     if c 
+                      
+                       case c 
+                       when 'A'
+                         p[0]+=1
+                         q[0]+=cq-33 if cq 
+                       when 'C'
+                         p[1]+=1
+                         q[1]+=cq-33 if cq 
+                       when 'G'
+                         p[2]+=1
+                         q[2]+=cq-33 if cq 
+                       when 'T'
+                         p[3]+=1
+                         q[3]+=cq-33 if cq 
+                       end  
+                      
+                       s+=c                
+                       
+                       print "#{c}" if show
+                   
+                     else
+                       del<<l
+                     end
+                   
+                   
+                   end
+                 
+                 
+                  
+                   pmax = p.index(p.max)
+                   max = p[pmax]
+                   p[pmax] = 0 if pmax
+                   palt = p.index(p.max)
+                   alt = p[palt]
+                   dep = max+alt
+                    max_q = q[pmax].to_f/max
+                    alt_q = '0'
+                    alt_q = q[palt].to_f/alt if alt > 0
+                   
+                     
+                   if dep > 2 and ((indel = indels_map[start] and indel[:count]>1 )  or  (alt>0 and alt/dep.to_f > 0.1 and alt_q >10) or alleles[ref]!=pmax )
+                     
+                     
+                         
+                     text = "#{start}\t#{max}/#{alt.to_i}\t#{ref}\t#{s}"  
+                     puts text  if var_show
+                     sum = (max+alt).to_f
+                     aalt = '-' 
+                     
+                     aalt = inv_alleles[palt] if alt!=0
+                     if (alt>0 and alt/dep.to_f > 0.1)
+                       c = 'a'
+                     else
+                       c = 'A'
+                     end
+                     
+                     
+                     if indel 
+                      s = indel[:alleles].split('/')
+                      ref = s[0]
+                      aalt = s[1]
+                      alt = indel[:count]
+                      if indel[:type]=='ins'
+                        c='+'
+                      else
+                        c='-'
+                      end
+                     end
+                     snps << [chr,start,ref,inv_alleles[pmax],aalt,max,alt,format('%.2f',max_q),format('%.2f',alt_q),c]
+                     
+                     # snps << [chr,start,ref,inv_alleles[pmax],inv_alleles[palt],max,alt,max/sum,alt/sum]
+                       
+                   end
+                   list-=del
+                   if list.size==0
+                   start = i.pos
+                   else
+                   start+=1
+                   end
+
+                 
+                   puts if show
+                 end
+               
+                 start = i.pos
+               end
+            
+         
+               
+               seq_size = i.seq.size
+               
+               tseq = form(i.seq,i.cigar)
+               tqual = form(i.qual,i.cigar)
+               
+               
+               if (i.cigar.index('I') or i.cigar.index('D') ) 
+
+                  
+                  if indel_show  
+                  puts "-------------------------INDEL " 
+                  puts i.pos
+                  puts i.cigar
+                  puts i.seq
+                  puts i.qual
+                  # puts i.inspect
+
+                  # :qname, :flag, :rname,:pos,:mapq,:cigar, :mrnm, :mpos, :isize, :seq, :qual, :tags, :al, :samstr
+                  
+                  puts seq[i.pos-chr_start..i.pos-chr_start+i.seq.size]
+                  puts tseq[0]
+                  puts tqual[0]
+                  puts tseq[1]
+                  end
+                  
+                            if tseq[1].size>0
+                              for y in tseq[1]
+                                if y[:type]=='ins'
+                                  iref = '-'
+                                  iseq = i.seq[y[:pos]..y[:pos]+y[:size]-1]
+                                  y[:alleles] =  "#{iref}/#{iseq}"
+                                else
+                                  iref = seq[i.pos-chr_start+y[:pos]..i.pos-chr_start+y[:pos]+y[:size]-1]
+                                  iseq = '-'
+                                  y[:alleles] =  "#{iref}/#{iseq}"
+                                end
+                                y[:pos]+=i.pos
+                                indels<<y
+                                unless indels_map[y[:pos]]
+                                    y[:count]=1
+                                    y[:sm]=y[:alleles]
+                                    indels_map[y[:pos]] = y
+                                else
+                                    indels_map[y[:pos]][:sm]+=",#{y[:alleles]}"
+                                  
+                                    indels_map[y[:pos]][:count] +=1
+                                end
+
+                              end
+                            end
+                  
+                  
+                end  
+                  
+                  
+        
+                    
+              
+              i.seq = tseq[0]
+              i.qual = tqual[0].bytes.to_a
+              
+              
+               if i.seq.size > seq_size/2
+              
+                   si = 0 
+               
+                   i.seq.size.times do |k|
+                     # puts "#{i.seq[k]} #{seq[start-chr_start+k]}"
+                     si+=1 if i.seq[k]==upseq[start-chr_start+k]
+                   end
+                   # puts "#{si} #{i.seq.size}"
+                   if (si.to_f/i.seq.size)>=0.95
+                     list << i 
+                   else
+                     if indel_show
+                     puts "Drop with low similarity #{si.to_f/i.seq.size} #{i.cigar}"
+                     puts seq[start-chr_start..start-chr_start+100]
+                     puts i.seq
+                     end
+                   end
+               end
+                                   
+               l = i 
+         
+    
+           }
+         
+          
+            if indel_show  
+ 
+                for i in indels_map.keys.sort
+                        puts indels_map[i].inspect
+                end
+           end
+           # puts l.cigar
+           # puts l.seq
+           # puts snps.size
+           return snps
+  
+end
+
+def self.call_variance p
+
+  
+  chr = p[:chr] 
+  
+  output = p[:output]
+  testBAMFile = p[:sam]
+  testReference = p[:ref]
+  sam = Bio::DB::Sam.new({:bam=>testBAMFile, :fasta=>testReference})
+  sam.open
+
+  chr_start = p[:start]
+  chr_stop  = p[:stop]
+  snps = call_sam_snv sam,chr,chr_start,chr_stop, p[:debug]  
+  
+  sam.close
+  f = File.open(output,'a')
+  f.puts snps.collect{|j| j.join("\t")}
+  f.close
+end
+
+
+
+def self.inspect_reference testReference
+  
+  # index reference 
+
+  file = File.open(testReference,'r')
+  map = {}
+  size=0
+  chr = nil
+  while str=file.gets
+    if str[0..0]=='>'
+      if chr  
+        puts "#{chr}\t#{size}"
+        map[chr]=size  
+      end 
+      chr = str.strip.split[0][1..-1]
+      size=0
+    else
+      size+=str.strip.size
+    end
+  end
+  map[chr]=size
+  puts map.inspect
+  file.close
+  genome = map
+  # genome = {"chr1"=>249250621, "chr10"=>135534747, "chr11"=>135006516, "chr12"=>133851895, "chr13"=>115169878, "chr14"=>107349540, "chr15"=>102531392, "chr16"=>90354753, "chr17"=>81195210, "chr18"=>78077248, "chr19"=>59128983, "chr2"=>243199373, "chr20"=>63025520, "chr21"=>48129895, "chr22"=>51304566, "chr3"=>198022430, "chr4"=>191154276, "chr5"=>180915260, "chr6"=>171115067, "chr7"=>159138663, "chr8"=>146364022, "chr9"=>141213431, "chrX"=>155270560, "chrY"=>59373566}
+
+  return genome
+end
+
+def self.inspect_reference_human testReference
+  
+  # index reference 
+  # 
+  # file = File.open(testReference,'r')
+  # map = {}
+  # size=0
+  # chr = nil
+  # while str=file.gets
+  #   if str[0..0]=='>'
+  #     if chr  
+  #       puts "#{chr}\t#{size}"
+  #       map[chr]=size  
+  #     end 
+  #     chr = str[1..-1].strip
+  #     size=0
+  #   else
+  #     size+=str.strip.size
+  #   end
+  # end
+  # map[chr]=size
+  # puts map.inspect
+  # file.close
+
+  genome = {"chr1"=>249250621, "chr10"=>135534747, "chr11"=>135006516, "chr12"=>133851895, "chr13"=>115169878, "chr14"=>107349540, "chr15"=>102531392, "chr16"=>90354753, "chr17"=>81195210, "chr18"=>78077248, "chr19"=>59128983, "chr2"=>243199373, "chr20"=>63025520, "chr21"=>48129895, "chr22"=>51304566, "chr3"=>198022430, "chr4"=>191154276, "chr5"=>180915260, "chr6"=>171115067, "chr7"=>159138663, "chr8"=>146364022, "chr9"=>141213431, "chrX"=>155270560, "chrY"=>59373566}
+
+  return genome
+end
+
+
+def self.process_vc_multicore  params, core=1,chr_filter=nil
+  
+  sim = core
+   
+  path = '/Users/soup/Desktop/vcqa'
+  output = params[:output]
+  
+  testReference = params[:ref]
+
+  genome = inspect_reference_human(testReference)
+
+  if chr_filter # 1 as text
+    
+      tmp = {}
+      chr = "chr#{chr_filter}"
+      tmp[chr]=genome[chr]
+      genome = tmp
+  end
+
+  puts genome.inspect
+
+
+  # reset output
+  f = File.open(output,'w')
+  f.close
+
+  a = []
+
+  # job generator
+  genome.each_pair do |chr,size|
+    puts "#{chr}\t#{size}"
+    m = 1000000
+    n = size/m
+
+    n.times do |i|
+      params[:chr] = chr
+      params[:start] = 1 + i * m
+      params[:stop] = (i+1) * m
+      a<<params.clone 
+    end
+
+    if size%m !=0
+      params[:chr] = chr
+      params[:start] = size - size% m
+      params[:stop] = size
+      a<<params.clone 
+    end
+
+  end
+
+  sim = a.size if sim>a.size
+
+  # starting first N processes
+  sim.times do
+    i = a.pop
+    Process.fork{call_variance(i)}
+  end
+  # start one by one as the previous finish
+  a.each do |i|
+    Process.wait(0)
+    Process.fork{call_variance(i)}
+  end
+  # wait for all to finish
+  Process.waitall
+  
+end
+
+
+
+def self.call_snv params
+  # ====================================================================================================
+  # 1. VC calling 
+  # ====================================================================================================
+  core = 4
+  core = params['core'].to_i if params['core'] # number of simultaneous processes
+  sam_file =params['sam']
+  ref_file =params['ref']
+  vc_file =params['vc_out']
+  p = {:sam=>sam_file,:ref=>ref_file,:output=>vc_file}
+
+  # params[:start] = 95_658_000
+  # params[:stop] = 95_659_000
+  # call_variance params
+  
+  process_vc_multicore p,core
+  # # params = {:sam=>sam_file,:ref=>ref_file,:output=>vc_all_file}
+  # # process_vc_multicore params,core
+end
+
+
+def self.call_snv_range params
+  # ====================================================================================================
+  # 1. VC calling 
+  # ====================================================================================================
+  core = 4
+  core = params['core'].to_i if params['core'] # number of simultaneous processes
+  sam_file =params['sam']
+  ref_file =params['ref']
+  vc_file =params['vc_out']
+  p = {:sam=>sam_file,:ref=>ref_file,:output=>vc_file,:chr=>"chr#{params['chr']}",:start=>params['start'].to_i,:stop=>params['stop'].to_i,:debug=>'indel'}
+  puts p.inspect
+  
+  call_variance p
+  
+end
+
+end
+end

data/lib/vardetect-vc/lib.rb

@@ -0,0 +1,28 @@
+require 'vardetect-vc/call_snv'
+require 'vardetect-vc/call_prepare'
+require 'vardetect-vc/call_combine'
+
+module Vardetect
+module Vc
+
+
+def self.parse_params params
+
+  puts params 
+  params = params.split
+  cmd = params[0]
+  tmp = {}
+  ((params.size-1)/2).times do |i|
+    key = params[1+i*2][1..-1]
+    value = params[1+i*2+1]
+    tmp[key]=value
+  end
+  tmp[:cmd]=cmd
+  params = tmp 
+end
+
+end
+
+end
+
+

data/lib/vardetect-vc/version.rb

@@ -1,5 +1,5 @@
 module Vardetect
   module Vc
-    VERSION = "0.0.4"
+    VERSION = "0.0.5"
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: vardetect-vc
 version: !ruby/object:Gem::Version
-  version: 0.0.4
+  version: 0.0.5
   prerelease: 
 platform: ruby
 authors:
@@ -78,7 +78,9 @@ dependencies:
 description: A variance calling tool for NGS
 email:
 - supasak.kul@biotec.or.th
-executables: []
+executables:
+- vardetect_vc
+- vardetect_vc_batch
 extensions: []
 extra_rdoc_files: []
 files:
@@ -87,7 +89,13 @@ files:
 - LICENSE.txt
 - README.md
 - Rakefile
+- bin/vardetect_vc
+- bin/vardetect_vc_batch
 - lib/vardetect-vc.rb
+- lib/vardetect-vc/call_combine.rb
+- lib/vardetect-vc/call_prepare.rb
+- lib/vardetect-vc/call_snv.rb
+- lib/vardetect-vc/lib.rb
 - lib/vardetect-vc/version.rb
 - vardetect-vc.gemspec
 homepage: ''