X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=code_ruby%2Flib%2Fmaasha%2Fsam.rb;h=8772e6382f20913d2fca3c9076b90475d2824564;hb=a80169a9121e8537f169cd85010d2ceae3a8d4fd;hp=d71862710baa4cb16e7d60874028882bec9fcdc0;hpb=7c18441aa7cd1049632fffc6b70256c0f1f64999;p=biopieces.git

diff --git a/code_ruby/lib/maasha/sam.rb b/code_ruby/lib/maasha/sam.rb
index d718627..8772e63 100644
--- a/code_ruby/lib/maasha/sam.rb
+++ b/code_ruby/lib/maasha/sam.rb
@@ -1,4 +1,4 @@
-# Copyright (C) 2007-2011 Martin A. Hansen.
+# Copyright (C) 2007-2013 Martin A. Hansen.
 
 # This program is free software; you can redistribute it and/or
 # modify it under the terms of the GNU General Public License
@@ -63,6 +63,7 @@ class Sam < Filesys
     bp[:STRAND]   = sam[:FLAG].revcomp? ? '-' : '+'
     bp[:S_ID]     = sam[:RNAME]
     bp[:S_BEG]    = sam[:POS]
+    bp[:S_END]    = sam[:POS] + sam[:SEQ].length - 1
     bp[:MAPQ]     = sam[:MAPQ]
     bp[:CIGAR]    = sam[:CIGAR].to_s
 
@@ -75,20 +76,44 @@ class Sam < Filesys
     bp[:SEQ]      = sam[:SEQ].seq
 
     unless sam[:SEQ].qual.nil?
-      bp[:SCORES] = sam[:SEQ].convert_phred2illumina!.qual
+      bp[:SCORES] = sam[:SEQ].qual_convert!(:base_33, :base_64).qual
     end
 
-    if sam.has_key? :NM and sam[:NM].to_i > 0
+    if sam[:NM] and sam[:NM].to_i > 0
+      bp[:NM] = sam[:NM]
+      bp[:MD] = sam[:MD]
       bp[:ALIGN] = self.align_descriptors(sam)
     end
 
     bp
   end
 
-  # Class method to convert a Biopiece record
-  # into a SAM entry.
-  def self.to_sam(bp)
-    "FISK" # FIXME
+  # Class method to create a new SAM entry from a Biopiece record.
+  # FIXME
+  def self.new_bp(bp)
+    qname = bp[:Q_ID]
+    flag  = 0
+    rname = bp[:S_ID]
+    pos   = bp[:S_BEG]
+    mapq  = bp[:MAPQ]
+    cigar = bp[:CIGAR]
+    rnext = bp[:Q_ID2]  || '*'
+    pnext = bp[:S_BEG2] || 0
+    tlen  = bp[:TLEN]   || 0
+    seq   = bp[:SEQ]
+    qual  = bp[:SCORES] || '*'
+    nm    = "NM:i:#{bp[:NM]}" if bp[:NM]
+    md    = "MD:Z:#{bp[:MD]}" if bp[:MD]
+
+    flag |= FLAG_REVCOMP if bp[:STRAND] == '+'
+
+    if qname && flag && rname && pos && mapq && cigar && rnext && pnext && tlen && seq && qual
+      ary = [qname, flag, rname, pos, mapq, cigar, rnext, pnext, tlen, seq, qual]
+      ary << nm if nm
+      ary << md if md
+      
+      ary.join("\t")
+    end
   end
 
   # Create alignment descriptors according to the KISS
@@ -239,7 +264,7 @@ class Sam < Filesys
 
     @header[:SQ][:SN] = Hash.new unless @header[:SQ][:SN].is_a? Hash
 
-    if @header[:SQ][:SN].has_key? seq_name
+    if @header[:SQ][:SN][seq_name]
       raise SamError, "Non-unique sequence name: #{seq_name}"
     else
       @header[:SQ][:SN][seq_name] = hash
@@ -267,13 +292,13 @@ class Sam < Filesys
       end
     end
 
-    if hash.has_key? :FO
+    if hash[:FO]
       unless hash[:FO] =~ /^\*|[ACMGRSVTWYHKDBN]+$/
         raise SamError, "Bad flow order: #{hash[:FO]}"
       end
     end
 
-    if hash.has_key? :PL
+    if hash[:PL]
       unless hash[:PL] =~ /^(CAPILLARY|LS454|ILLUMINA|SOLID|HELICOS|IONTORRENT|PACBIO)$/
         raise SamError, "Bad platform: #{hash[:PL]}"
       end
@@ -281,7 +306,7 @@ class Sam < Filesys
 
     @header[:RG][:ID] = Hash.new unless @header[:RG][:ID].is_a? Hash
 
-    if @header[:RG][:ID].has_key? id
+    if @header[:RG][:ID][id]
       raise SamError, "Non-unique read group identifier: #{id}"
     else
       @header[:RG][:ID][id] = hash
@@ -311,7 +336,7 @@ class Sam < Filesys
 
     @header[:PG][:ID] = Hash.new unless @header[:PG][:ID].is_a? Hash
 
-    if @header[:PG][:ID].has_key? id
+    if @header[:PG][:ID][id]
       raise SamError, "Non-unique program record identifier: #{id}"
     else
       @header[:PG][:ID][id] = hash
@@ -368,7 +393,7 @@ class Sam < Filesys
     entry[:RNEXT] = rnext
     entry[:PNEXT] = pnext
     entry[:TLEN]  = tlen
-    entry[:SEQ]   = (qual == '*') ? Seq.new(qname, seq) : Seq.new(qname, seq, qual)
+    entry[:SEQ]   = (qual == '*') ? Seq.new(seq_name: qname, seq: seq) : Seq.new(seq_name: qname, seq: seq, qual: qual)
     entry[:QUAL]  = qual
 
     # Optional fields - where some are really important! HATE HATE HATE SAM!!!
@@ -376,7 +401,7 @@ class Sam < Filesys
     fields[11 .. -1].each do |field|
       tag, type, val = field.split(':')
 
-      raise SamError, "Non-unique optional tag: #{tag}" if entry.has_key? tag.to_sym
+      raise SamError, "Non-unique optional tag: #{tag}" if entry[tag.to_sym]
 
       # A [!-~] Printable character
 
@@ -414,7 +439,7 @@ class Sam < Filesys
     raise SamError, "Bad rname: #{rname}" unless rname =~ /^(\*|[!-()+-<>-~][!-~]*)$/
 
     unless @header.empty? or rname == '*'
-      unless @header[:SQ][:SN].has_key? rname.to_sym
+      unless @header[:SQ][:SN][rname.to_sym]
         raise SamError, "rname not found in header hash: #{rname}"
       end
     end
@@ -437,7 +462,7 @@ class Sam < Filesys
     raise SamError, "Bad rnext: #{rnext}" unless rnext =~ /^(\*|=|[!-()+-<>-~][!-~]*)$/
 
     unless @header.empty? or rnext == '*' or rnext == '='
-      unless @header[:SQ][:SN].has_key? rnext.to_sym
+      unless @header[:SQ][:SN][rnext.to_sym]
         raise SamError, "rnext not found in header hash: #{rnext}"
       end
     end