htslib 0.2.6 → 0.2.9

data/lib/hts/bam/record.rb

@@ -12,9 +12,64 @@ module HTS
 
       attr_reader :header
 
-      def initialize(bam1_t, header)
-        @bam1 = bam1_t
+      # Initialization API is experimental.
+
+      def initialize(
+        header,
+        bam1 = nil,
+        qname: nil,
+        flag: nil,
+        tid: nil,
+        pos: nil,
+        mapq: nil,
+        cigar: nil,
+        mtid: nil,
+        mpos: nil,
+        isize: nil,
+        seq: nil,
+        qual: nil,
+        l_aux: nil
+      )
+        @bam1 = bam1 || LibHTS.bam_init1
         @header = header
+
+        params = [qname, flag, tid, pos, mapq, cigar, mtid, mpos, isize, seq, qual, l_aux]
+        return if params.all? { |x| x.nil? }
+
+        if params.all?
+          c = FFI::MemoryPointer.new(:pointer)
+          m = FFI::MemoryPointer.new(:size_t)
+          LibHTS.sam_parse_cigar(cigar, FFI::Pointer::NULL, c, m)
+          cigar_array = c.read_pointer.read_array_of_uint32(m.read(:size_t))
+          cigar_pointer = FFI::MemoryPointer.new(:uint32, cigar_array.length)
+          cigar_pointer.write_array_of_uint32(cigar_array)
+          if qual.is_a?(Array)
+            qual = qual.pack("C*")
+          elsif qual.is_a?(String)
+            raise "Which implementation is better? +33 or not? Please tell me."
+          end
+          r = LibHTS.bam_set1(
+            @bam1,
+            qname.length,
+            qname,
+            flag,
+            tid,
+            pos,
+            mapq,
+            cigar_array.length,
+            cigar_pointer,
+            mtid,
+            mpos,
+            isize,
+            seq.length,
+            seq,
+            qual,
+            l_aux
+          )
+          raise "bam_set1 failed: #{r}" if r < 0
+        else
+          warn "Ignore bam_set1 because some arguments are missing."
+        end
       end
 
       # Return the FFI::Struct object.
@@ -32,6 +87,10 @@ module HTS
         LibHTS.bam_get_qname(@bam1).read_string
       end
 
+      def qname=(name)
+        LibHTS.bam_set_qname(@bam1, name)
+      end
+
       # Get the chromosome ID of the alignment. -1 if not mapped.
       # @return [Integer] chromosome ID
       def tid
@@ -154,6 +213,17 @@ module HTS
         Cigar.new(self)
       end
 
+      def cigar=(str)
+        if cigar.is_a? Cigar
+          raise "Not implemented yet."
+        elsif cigar.is_a? String
+          r = LibHTS.bam_parse_cigar(str, FFI::Pointer::NULL, @bam1)
+          raise "bam_parse_cigar failed: #{r}" if r != 0
+        else
+          raise ArgumentError, "cigar must be a String or Bam::Cigar"
+        end
+      end
+
       # Calculate query length from CIGAR.
       # @return [Integer] query length
       def qlen
@@ -178,7 +248,7 @@ module HTS
       def seq
         r = LibHTS.bam_get_seq(@bam1)
         seq = String.new
-        (@bam1[:core][:l_qseq]).times do |i|
+        len.times do |i|
           seq << SEQ_NT16_STR[LibHTS.bam_seqi(r, i)]
         end
         seq
@@ -195,8 +265,8 @@ module HTS
       # @param [Integer] i index
       # @return [String] base
       def base(n)
-        n += @bam1[:core][:l_qseq] if n < 0
-        return "." if (n >= @bam1[:core][:l_qseq]) || (n < 0) # eg. base(-1000)
+        n += len if n < 0
+        return "." if (n >= len) || (n < 0) # eg. base(-1000)
 
         r = LibHTS.bam_get_seq(@bam1)
         SEQ_NT16_STR[LibHTS.bam_seqi(r, n)]
@@ -206,7 +276,7 @@ module HTS
       # @return [Array] base qualities
       def qual
         q_ptr = LibHTS.bam_get_qual(@bam1)
-        q_ptr.read_array_of_uint8(@bam1[:core][:l_qseq])
+        q_ptr.read_array_of_uint8(len)
       end
 
       # Get the base qualities as a string. (a.k.a QUAL)
@@ -220,8 +290,8 @@ module HTS
       # @param [Integer] i index
       # @return [Integer] base quality
       def base_qual(n)
-        n += @bam1[:core][:l_qseq] if n < 0
-        return 0 if (n >= @bam1[:core][:l_qseq]) || (n < 0) # eg. base_qual(-1000)
+        n += len if n < 0
+        return 0 if (n >= len) || (n < 0) # eg. base_qual(-1000)
 
         q_ptr = LibHTS.bam_get_qual(@bam1)
         q_ptr.get_uint8(n)
@@ -256,11 +326,11 @@ module HTS
         end
       end
 
-      # TODO: add a method to get the auxillary fields as a hash.
+      # TODO: add a method to get the auxiliary fields as a hash.
 
-      # TODO: add a method to set the auxillary fields.
+      # TODO: add a method to set the auxiliary fields.
 
-      # TODO: add a method to remove the auxillary fields.
+      # TODO: add a method to remove the auxiliary fields.
 
       # TODO: add a method to set variable length data (qname, cigar, seq, qual).
 

data/lib/hts/bam.rb

@@ -114,29 +114,12 @@ module HTS
     def write(record)
       check_closed
 
-      # record = record.dup
       r = LibHTS.sam_write1(@hts_file, header, record)
       raise "Failed to write record" if r < 0
     end
 
-    def <<(aln)
-      write(aln)
-    end
-
-    def each(copy: false, &block)
-      if copy
-        each_record_copy(&block)
-      else
-        each_record_reuse(&block)
-      end
-    end
-
-    def query(region, copy: false, &block)
-      if copy
-        query_copy(region, &block)
-      else
-        query_reuse(region, &block)
-      end
+    def <<(record)
+      write(record)
     end
 
     # @!macro [attach] define_getter
@@ -158,9 +141,10 @@ module HTS
     alias isize insert_size
     alias mpos mate_pos
 
+    # FXIME: experimental
     def aux(tag)
-      warn "experimental"
       check_closed
+
       position = tell
       ary = map { |r| r.aux(tag) }
       seek(position)
@@ -185,8 +169,8 @@ module HTS
     alias each_isize each_insert_size
     alias each_mpos each_mate_pos
 
+    # FIXME: experimental
     def each_aux(tag)
-      warn "experimental"
       check_closed
       return to_enum(__method__, tag) unless block_given?
 
@@ -197,47 +181,44 @@ module HTS
       self
     end
 
-    private
+    def each(copy: false, &block)
+      if copy
+        each_record_copy(&block)
+      else
+        each_record_reuse(&block)
+      end
+    end
 
-    def query_reuse(region)
+    def query(region, beg = nil, end_ = nil, copy: false, &block)
       check_closed
       raise "Index file is required to call the query method." unless index_loaded?
-      return to_enum(__method__, region) unless block_given?
-
-      qiter = LibHTS.sam_itr_querys(@idx, header, region)
-      raise "Failed to query region: #{region}" if qiter.null?
 
-      bam1 = LibHTS.bam_init1
-      record = Record.new(bam1, header)
-      begin
-        yield record while LibHTS.sam_itr_next(@hts_file, qiter, bam1) > 0
-      ensure
-        LibHTS.hts_itr_destroy(qiter)
+      if beg && end_
+        tid = header.get_tid(region)
+        queryi(tid, beg, end_, copy:, &block)
+      elsif beg.nil? && end_.nil?
+        querys(region, copy:, &block)
+      else
+        raise ArgumentError, "beg and end_ must be specified together"
       end
-      self
     end
 
-    def query_copy(region)
-      check_closed
-      raise "Index file is required to call the query method." unless index_loaded?
-      return to_enum(__method__, region) unless block_given?
+    private
 
-      qiter = LibHTS.sam_itr_querys(@idx, header, region)
-      raise "Failed to query region: #{region}" if qiter.null?
+    def queryi(tid, beg, end_, copy: false, &block)
+      if copy
+        queryi_copy(tid, beg, end_, &block)
+      else
+        queryi_reuse(tid, beg, end_, &block)
+      end
+    end
 
-      begin
-        loop do
-          bam1 = LibHTS.bam_init1
-          slen = LibHTS.sam_itr_next(@hts_file, qiter, bam1)
-          break if slen == -1
-          raise if slen < -1
-
-          yield Record.new(bam1, header)
-        end
-      ensure
-        LibHTS.hts_itr_destroy(qiter)
+    def querys(region, copy: false, &block)
+      if copy
+        querys_copy(region, &block)
+      else
+        querys_reuse(region, &block)
       end
-      self
     end
 
     def each_record_reuse
@@ -247,7 +228,7 @@ module HTS
       return to_enum(__method__) unless block_given?
 
       bam1 = LibHTS.bam_init1
-      record = Record.new(bam1, header)
+      record = Record.new(header, bam1)
       yield record while LibHTS.sam_read1(@hts_file, header, bam1) != -1
       self
     end
@@ -257,10 +238,73 @@ module HTS
       return to_enum(__method__) unless block_given?
 
       while LibHTS.sam_read1(@hts_file, header, bam1 = LibHTS.bam_init1) != -1
-        record = Record.new(bam1, header)
+        record = Record.new(header, bam1)
         yield record
       end
       self
     end
+
+    def queryi_reuse(tid, beg, end_, &block)
+      return to_enum(__method__, region, beg, end_) unless block_given?
+
+      qiter = LibHTS.sam_itr_queryi(@idx, tid, beg, end_)
+      raise "Failed to query region: #{tid} #{beg} #{end_}" if qiter.null?
+
+      query_reuse_yield(qiter, &block)
+      self
+    end
+
+    def queryi_copy(tid, beg, end_, &block)
+      return to_enum(__method__, tid, beg, end_) unless block_given?
+
+      qiter = LibHTS.sam_itr_queryi(@idx, tid, beg, end_)
+      raise "Failed to query region: #{tid} #{beg} #{end_}" if qiter.null?
+
+      query_copy(qiter, &block)
+      self
+    end
+
+    def querys_reuse(region, &block)
+      return to_enum(__method__, region) unless block_given?
+
+      qiter = LibHTS.sam_itr_querys(@idx, header, region)
+      raise "Failed to query region: #{region}" if qiter.null?
+
+      query_reuse_yield(qiter, &block)
+      self
+    end
+
+    def querys_copy(region, &block)
+      return to_enum(__method__, region) unless block_given?
+
+      qiter = LibHTS.sam_itr_querys(@idx, header, region)
+      raise "Failed to query region: #{region}" if qiter.null?
+
+      query_copy(qiter, &block)
+      self
+    end
+
+    def query_reuse_yield(qiter)
+      bam1 = LibHTS.bam_init1
+      record = Record.new(header, bam1)
+      begin
+        yield record while LibHTS.sam_itr_next(@hts_file, qiter, bam1) > 0
+      ensure
+        LibHTS.hts_itr_destroy(qiter)
+      end
+    end
+
+    def query_copy(qiter)
+      loop do
+        bam1 = LibHTS.bam_init1
+        slen = LibHTS.sam_itr_next(@hts_file, qiter, bam1)
+        break if slen == -1
+        raise if slen < -1
+
+        yield Record.new(header, bam1)
+      end
+    ensure
+      LibHTS.hts_itr_destroy(qiter)
+    end
   end
 end

data/lib/hts/bcf/header.rb

@@ -19,6 +19,8 @@ module HTS
         else
           raise TypeError, "Invalid argument"
         end
+
+        yield self if block_given?
       end
 
       def struct
@@ -94,6 +96,14 @@ module HTS
         kstr[:s]
       end
 
+      def name2id(name)
+        LibHTS.bcf_hdr_name2id(@bcf_hdr, name)
+      end
+
+      def id2name(id)
+        LibHTS.bcf_hdr_id2name(@bcf_hdr, id)
+      end
+
       private
 
       def bcf_hl_type_to_int(bcf_hl_type)

data/lib/hts/bcf/record.rb

@@ -4,8 +4,8 @@ module HTS
   class Bcf < Hts
     # A class for working with VCF records.
     class Record
-      def initialize(bcf_t, header)
-        @bcf1 = bcf_t
+      def initialize(header, bcf_t = nil)
+        @bcf1 = bcf_t || LibHTS.bcf_init
         @header = header
       end
 

data/lib/hts/bcf.rb

@@ -63,7 +63,7 @@ module HTS
         warn "Create index for #{@file_name}"
       end
       case LibHTS.bcf_index_build3(@file_name, index_name, min_shift, (@nthreads || threads))
-      when 0 # sccessful
+      when 0 # successful
       when -1 then raise "indexing failed"
       when -2 then raise "opening #{@file_name} failed"
       when -3 then raise "format not indexable"
@@ -98,7 +98,7 @@ module HTS
     def write_header(header)
       check_closed
 
-      @header = header.dup
+      @header = header.clone
       LibHTS.bcf_hdr_write(@hts_file, header)
     end
 
@@ -130,35 +130,6 @@ module HTS
       header.samples
     end
 
-    def each(copy: false, &block)
-      if copy
-        each_record_copy(&block)
-      else
-        each_record_reuse(&block)
-      end
-    end
-
-    def query(...)
-      querys(...) # Fixme
-    end
-
-    # def queryi
-    # end
-
-    def querys(region, copy: false, &block)
-      if copy
-        querys_copy(region, &block)
-      else
-        querys_reuse(region, &block)
-      end
-    end
-
-    # private def queryi_copy
-    # end
-
-    # private def queryi_reuse
-    # end
-
     # @!macro [attach] define_getter
     #   @method $1
     #   Get $1 array
@@ -214,7 +185,7 @@ module HTS
 
     def each_info(key)
       check_closed
-      return to_enum(__method__, key) unless block
+      return to_enum(__method__, key) unless block_given?
 
       each do |r|
         yield r.info(key)
@@ -223,27 +194,78 @@ module HTS
 
     def each_format(key)
       check_closed
-      return to_enum(__method__, key) unless block
+      return to_enum(__method__, key) unless block_given?
 
       each do |r|
         yield r.format(key)
       end
     end
 
-    private
+    def each(copy: false, &block)
+      if copy
+        each_record_copy(&block)
+      else
+        each_record_reuse(&block)
+      end
+    end
 
-    def querys_reuse(region)
+    def query(region, beg = nil, end_ = nil, copy: false, &block)
       check_closed
 
       raise "query is only available for BCF files" unless file_format == "bcf"
       raise "Index file is required to call the query method." unless index_loaded?
+
+      if beg && end_
+        tid = header.name2id(region)
+        queryi(tid, beg, end_, copy:, &block)
+      elsif beg.nil? && end_.nil?
+        querys(region, copy:, &block)
+      else
+        raise ArgumentError, "beg and end must be specified together"
+      end
+    end
+
+    private
+
+    def queryi(tid, beg, end_, copy: false, &block)
+      if copy
+        queryi_copy(tid, beg, end_, &block)
+      else
+        queryi_reuse(tid, beg, end_, &block)
+      end
+    end
+
+    def querys(region, copy: false, &block)
+      if copy
+        querys_copy(region, &block)
+      else
+        querys_reuse(region, &block)
+      end
+    end
+
+    def queryi_reuse(tid, beg, end_, &block)
+      return to_enum(__method__, tid, beg, end_) unless block_given?
+
+      qiter = LibHTS.bcf_itr_queryi(@idx, tid, beg, end_)
+      raise "Failed to query region #{tid} #{beg} #{end_}" if qiter.null?
+
+      query_reuse_yield(qiter, &block)
+      self
+    end
+
+    def querys_reuse(region, &block)
       return to_enum(__method__, region) unless block_given?
 
       qiter = LibHTS.bcf_itr_querys(@idx, header, region)
       raise "Failed to query region #{region}" if qiter.null?
 
+      query_reuse_yield(qiter, &block)
+      self
+    end
+
+    def query_reuse_yield(qiter)
       bcf1 = LibHTS.bcf_init
-      record = Record.new(bcf1, header)
+      record = Record.new(header, bcf1)
       begin
         loop do
           slen = LibHTS.hts_itr_next(@hts_file[:fp][:bgzf], qiter, bcf1, ::FFI::Pointer::NULL)
@@ -255,32 +277,39 @@ module HTS
       ensure
         LibHTS.bcf_itr_destroy(qiter)
       end
-      self
     end
 
-    def querys_copy(region)
-      check_closed
+    def queryi_copy(tid, beg, end_, &block)
+      return to_enum(__method__, tid, beg, end_) unless block_given?
 
-      raise "query is only available for BCF files" unless file_format == "bcf"
-      raise "Index file is required to call the query method." unless index_loaded?
+      qiter = LibHTS.bcf_itr_queryi(@idx, tid, beg, end_)
+      raise "Failed to query region #{tid} #{beg} #{end_}" if qiter.null?
+
+      query_copy_yield(qiter, &block)
+      self
+    end
+
+    def querys_copy(region, &block)
       return to_enum(__method__, region) unless block_given?
 
       qiter = LibHTS.bcf_itr_querys(@idx, header, region)
       raise "Failed to query region #{region}" if qiter.null?
 
-      begin
-        loop do
-          bcf1 = LibHTS.bcf_init
-          slen = LibHTS.hts_itr_next(@hts_file[:fp][:bgzf], qiter, bcf1, ::FFI::Pointer::NULL)
-          break if slen == -1
-          raise if slen < -1
+      query_copy_yield(qiter, &block)
+      self
+    end
 
-          yield Record.new(bcf1, header)
-        end
-      ensure
-        LibHTS.bcf_itr_destroy(qiter)
+    def query_copy_yield(qiter)
+      loop do
+        bcf1 = LibHTS.bcf_init
+        slen = LibHTS.hts_itr_next(@hts_file[:fp][:bgzf], qiter, bcf1, ::FFI::Pointer::NULL)
+        break if slen == -1
+        raise if slen < -1
+
+        yield Record.new(header, bcf1)
       end
-      self
+    ensure
+      LibHTS.bcf_itr_destroy(qiter)
     end
 
     def each_record_reuse
@@ -289,7 +318,7 @@ module HTS
       return to_enum(__method__) unless block_given?
 
       bcf1 = LibHTS.bcf_init
-      record = Record.new(bcf1, header)
+      record = Record.new(header, bcf1)
       yield record while LibHTS.bcf_read(@hts_file, header, bcf1) != -1
       self
     end
@@ -300,7 +329,7 @@ module HTS
       return to_enum(__method__) unless block_given?
 
       while LibHTS.bcf_read(@hts_file, header, bcf1 = LibHTS.bcf_init) != -1
-        record = Record.new(bcf1, header)
+        record = Record.new(header, bcf1)
         yield record
       end
       self

data/lib/hts/faidx/sequence.rb

@@ -1,5 +1,3 @@
-require_relative "../faidx"
-
 module HTS
   class Faidx
     class Sequence

data/lib/hts/faidx.rb

@@ -26,7 +26,8 @@ module HTS
       end
 
       @file_name = file_name
-      @fai = if [".fq", ".fastq"].include? File.extname(@file_name)
+      @fai = case File.extname(@file_name)
+             when ".fq", ".fastq"
                LibHTS.fai_load_format(@file_name, 2)
              else
                LibHTS.fai_load(@file_name)
@@ -40,7 +41,14 @@ module HTS
     end
 
     def close
+      return if closed?
+
       LibHTS.fai_destroy(@fai)
+      @fai = nil
+    end
+
+    def closed?
+      @fai.nil? || @fai.null?
     end
 
     def file_format
@@ -100,14 +108,14 @@ module HTS
       rlen = FFI::MemoryPointer.new(:int)
 
       if start.nil? && stop.nil?
-        result = LibHTS.fai_fetch(@fai, name, rlen)
+        result = LibHTS.fai_fetch64(@fai, name, rlen)
       else
         start < 0    && raise(ArgumentError, "Expect start to be >= 0")
         stop  < 0    && raise(ArgumentError, "Expect stop to be >= 0")
         start > stop && raise(ArgumentError, "Expect start to be <= stop")
         stop >= seq_len(name) && raise(ArgumentError, "Expect stop to be < seq_len")
 
-        result = LibHTS.faidx_fetch_seq(@fai, name, start, stop, rlen)
+        result = LibHTS.faidx_fetch_seq64(@fai, name, start, stop, rlen)
       end
 
       case rlen.read_int
@@ -125,14 +133,14 @@ module HTS
       rlen = FFI::MemoryPointer.new(:int)
 
       if start.nil? && stop.nil?
-        result = LibHTS.fai_fetchqual(@fai, name, rlen)
+        result = LibHTS.fai_fetchqual64(@fai, name, rlen)
       else
         start < 0    && raise(ArgumentError, "Expect start to be >= 0")
         stop  < 0    && raise(ArgumentError, "Expect stop to be >= 0")
         start > stop && raise(ArgumentError, "Expect start to be <= stop")
         stop >= seq_len(name) && raise(ArgumentError, "Expect stop to be < seq_len")
 
-        result = LibHTS.faidx_fetch_qual(@fai, name, start, stop, rlen)
+        result = LibHTS.faidx_fetch_qual64(@fai, name, start, stop, rlen)
       end
 
       case rlen.read_int

data/lib/hts/libhts/bgzf.rb

@@ -41,7 +41,7 @@ module HTS
       :ssize_t
 
     # Write _length_ bytes from _data_ to the file, the index will be used to
-    # decide the amount of uncompressed data to be writen to each bgzip block.
+    # decide the amount of uncompressed data to be written to each bgzip block.
     attach_function \
       :bgzf_block_write,
       [BGZF, :pointer, :size_t],