libssw 0.0.0.pre → 0.0.1.pre

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: f52642bad07253a1fa4ad73032c49d805a53080812bbb663c0fa825c6b0241f0
-  data.tar.gz: af7af5c499367ad64f4914bd1c4efce747966170d181960fa20b2c419745666b
+  metadata.gz: a45ea25e49633d46da6d363c9ed1d23c9fd3ed1a07b6128d4d61e4b320a7ff44
+  data.tar.gz: 599aee0f039c17fe6b3b50e4e89302e2a642b9a6ec886b0dd6ebc13e8972387b
 SHA512:
-  metadata.gz: 3858b66d605d0011ce6e9c6606befbd7520613a1435297276a707cf22431ee38e043c7402ac336a08851da79ffd097bf28d72cd8debd995e66e1c1466c3a48b5
-  data.tar.gz: 7aed161e8a3a8d8c9b754749af60408dbdcc05a8076efe3d116d76fdccac03dbee967cf95a6c7f042cdf7b9cdb99b24e9f059e995c351079a8a156316198f3e7
+  metadata.gz: b03f65581c285cedb488946ae84fd029147b54a6b022b9aff5bf39f7554881d8eb2e8a60064f06f012abe7fee21b2aff1801ed1017d7c3ef995557a7a066b687
+  data.tar.gz: e651c5fc0c9fb918a479c5070a00631e308562191d7adb0784e7042d07874e5026245c4c76acb48cf99900739e6c0da2617be2100398dbd1134b011a565a0014

data/exe/rbssw

@@ -1,4 +1,213 @@
 #!/usr/bin/env ruby
 # frozen_string_literal: true
 
+require 'bio'
 require 'libssw'
+SSW = LibSSW
+require 'optparse'
+
+opts = {
+  lib_path: nil,
+  nmatch: 2,
+  nmismatch: 2,
+  nopen: 3,
+  next: 1,
+  bprotein: false,
+  smatrix: nil,
+  bpath: false,
+  nthr: nil,
+  bbest: false,
+  bsam: nil, # typo?
+  bheader: nil
+}
+
+parser = OptionParser.new do |opt|
+  opt.version = LibSSW::VERSION
+  opt.summary_width = 20
+  opt.banner = 'Usage: rbssw [options] <target file> <query file>'
+  opt.on('-l', '--sLibPath PATH', String,
+         'path of libssw.so') do |v|
+           opts[:sLibPath] = v
+         end
+  opt.on('-m', '--nMatch VAL', Integer,
+         'a positive integer as the score for a match',
+         "in genome sequence alignment. [#{opts[:nmatch]}]") do |v|
+           opts[:nmatch] = v
+         end
+  opt.on('-x', '--nMismatch VAL', Integer,
+         'a positive integer as the score for a mismatch',
+         "in genome sequence alignment. [#{opts[:nmismatch]}]") do |v|
+           opts[:nmismatch] = v
+         end
+  opt.on('-o', '--nOpen VAL', Integer,
+         'a positive integer as the penalty for the gap opening',
+         "in genome sequence alignment. [#{opts[:nopen]}]") do |v|
+           opts[:nopen] = v
+         end
+  opt.on('-e', '--nExt VAL', Integer,
+         'a positive integer as the penalty for the gap extension',
+         "in genome sequence alignment. [#{opts[:next]}]") do |v|
+           opts[:next] = v
+         end
+  opt.on('-p', '--bProtien', TrueClass,
+         'Do protein sequence alignment.',
+         "Without this option, do genome sequence alignment. [#{opts[:bprotein]}]") do |v|
+           opts[:bprotein] = v
+         end
+  opt.on('-a', '--sMatrix VAL', String,
+         "a file for either Blosum or Pam weight matrix. [#{opts[:smatrix]}]") do |v|
+           opts[:smatrix] = v
+         end
+  opt.on('-c', '--bPath', TrueClass,
+         "Return the alignment path. [#{opts[:bpath]}]") do |v|
+           opts[:bpath] = v
+         end
+  opt.on('-f', '--nThr VAL', Integer,
+         'a positive integer.',
+         'Only output the alignments with the Smith-Waterman score >= N.') do |v|
+           opts[:nthr] = v
+         end
+  opt.on('-r', '--bBest', TrueClass,
+         'The best alignment will be picked, between the original read',
+         "alignment and the reverse complement read alignment. [#{opts[:bbest]}]") do |v|
+           opts[:bbest] = v
+         end
+  opt.on('-s', '--bSam', TrueClass,
+         'Output in SAM format. [no header]') do |v| # TYPO?
+           opts[:bsam] = v
+         end
+  opt.on('-header', '--bHeader', TrueClass,
+         'If -s is used, include header in SAM output.') do |v|
+           opts[:bheader] = v
+         end
+end
+
+parser.order!(ARGV)
+
+opts[:target] = ARGV[0]
+opts[:query]  = ARGV[1]
+
+lEle = []
+dRc = {}
+dEle2Int = {}
+dInt2Ele = {}
+lScore = nil
+
+if opts[:bprotein]
+  # load AA score matrix
+  if !opts[:smatrix]
+    lEle = ['A', 'R', 'N', 'D', 'C', 'Q', 'E', 'G', 'H', 'I', 'L', 'K',
+            'M', 'F', 'P', 'S', 'T', 'W', 'Y', 'V', 'B', 'Z', 'X', '*']
+    lEle.each_with_index do |ele, i|
+      dEle2Int[ele] = i
+      dEle2Int[ele.downcase] = i
+      dInt2Ele[i] = ele
+    end
+    nEleNum = lEle.size
+    lScore = SSW::Blosum50
+  else
+    lEle, dEle2Int, dInt2Ele, lScore = SSW.read_matrix(opts[:smatrix])
+  end
+elsif !opts[:smatrix]
+  # init DNA score matrix
+  lEle = %w[A C G T N]
+  dRc = { 'A': 'C', 'C': 'G', 'G': 'C', 'T': 'A', 'a': 'C', 'c': 'G', 'g': 'C', 't': 'A' }
+  lEle.each_with_index do |ele, i|
+    dEle2Int[ele] = i
+    dEle2Int[ele.downcase] = i
+    dInt2Ele[i] = ele
+  end
+  # dEle2Int = {'A': 0, 'a': 0, 'C': 1, 'G': 2, 'g': 2, 'c': 1, 'N': 4, 'T': 3, 'n': 4, 't': 3}
+  # dInt2Ele = {0: 'A', 1: 'C', 2: 'G', 3: 'T', 4: 'N'}
+  nEleNum = lEle.size # 5
+  lScore = Array.new(nEleNum**2, 0)
+  (nEleNum - 1).times do |i|
+    (nEleNum - 1).times do |j|
+      lScore[i * nEleNum + j] = if lEle[i] == lEle[j]
+                                  opts[:nmatch]
+                                else
+                                  -opts[:nmismatch]
+                                end
+    end
+  end
+  # lScore = [ 2, -2, -2, -2, 0,
+  #           -2,  2, -2, -2, 0,
+  #           -2, -2,  2, -2, 0,
+  #           -2, -2, -2,  2, 0,
+  #            0,  0,  0,  0, 0 ]
+end
+
+warn 'Reverse complement alignment is not available for protein sequences.' if opts[:bbest] && opts[:bprotein]
+
+# set flag
+nFlag = opts[:bpath] ? 2 : 0
+
+# print sam head
+if opts[:bsam] && opts[:bheader] && opts[:bpath]
+  puts '@HD\tVN:1.4\tSO:queryname'
+  Bio::Flatfile.open(opts[:target]) do |f|
+    f.each do |entry|
+      id  = entry.entry_id
+      len = entry.nalen
+      puts "@SQ\tSN:#{id}\tLN:#{len}"
+    end
+  end
+elsif opts[:bsam] && !args[:bpath]
+  warn 'SAM format output is only available together with option -c.\n'
+  opts[:bsam] = false
+end
+
+def _to_int(seq, lEle, dEle2Int)
+  seq.each_char.map do |ele|
+    if dEle2Int.has_key?(ele)
+      dEle2Int[ele]
+    else
+      dEle2Int[lEle[-1]]
+    end
+  end
+end
+
+# iterate query sequenc
+Bio::FlatFile.open(opts[:query]) do |query_file|
+  query_file.each do |qentry|
+    sQId   = qentry.entry_id
+    sQSeq  = qentry.sequence_string
+    sQQual = qentry.quality_string
+    # build query profile
+    qNum = _to_int(sQSeq, lEle, dEle2Int)
+    qProfile = SSW.ssw_init(qNum, sQSeq.size, lScore, lEle.size, 2)
+    # build rc query profile
+    if opts[:bbest] && !opts[:bprotein]
+      sQRcSeq = sQSeq.reverse.each_char.map { |x| dRc[x] }.join
+      qRcNum = _to_int(sQRcSeq, lEle, dEle2Int)
+      qRcProfile = SSW.ssw_init(qRcNum, sQSeq.size, mat, lEle.size, 2)
+    end
+    # set mask le
+    if sQSeq.size > 30
+      nMaskLen = sQSeq.size / 2
+    else
+      nMasklen = 15
+    end
+
+    # iter target sequence
+    Bio::FlatFile.open(opts[:target]) do |target_file|
+      target_file.each do |tentry|
+        sRId  = tentry.entry_id
+        sRSeq = tentry.seq.to_s
+        rNum  = _to_int(sRSeq, lEle, dEle2Int)
+        res = SSW.ssw_align(
+          qProfile, rNum, sRSeq.size, opts[:nopen], opts[:next], nFlag, 0, 0, nMaskLen
+        )
+        p res.to_h
+        resRc = nil
+        if opts[:bbest] && !opts[:bprotein]
+          resRc = SSW.align_one(
+            qRcProfile, rNum, sRSeq.size, opts[:nopen], opts[:next], nFlag, 0, 0, nMaskLen
+          )
+        end
+        # build cigar and trace back path
+        strand = 0
+      end
+    end
+  end
+end

data/lib/libssw.rb

@@ -1,6 +1,5 @@
 # frozen_string_literal: true
 
-require 'forwardable'
 require_relative 'libssw/version'
 
 module LibSSW
@@ -25,14 +24,97 @@ module LibSSW
                    File.expand_path("../vendor/#{lib_name}", __dir__)
                  end
 
-  autoload :FFI, 'libssw/ffi'
+  require_relative 'libssw/ffi'
 
-  extend Forwardable
-  Align   = FFI::Align
-  Profile = FFI::Profile
-  def_delegators :FFI,
-                 :ssw_init,
-                 :init_destroy,
-                 :align_destroy,
-                 :mark_mismatch
+  class Align < FFI::Align
+    def cigar
+      pt = super
+      return [] if cigar_len.zero?
+
+      pt[0, 4 * cigar_len].unpack('L*')
+    end
+
+    def cigar_len
+      cigarLen
+    end
+
+    def to_h
+      h = {}
+      %i[score1
+         score2
+         ref_begin1
+         ref_end1
+         read_begin1
+         read_end1
+         ref_end2
+         cigar
+         cigar_len].each do |k|
+        h[k] = __send__(k)
+      end
+      h
+    end
+  end
+
+  class Profile < FFI::Profile
+    def read
+      pt = super
+      return [] if read_len.zero?
+
+      pt[0, read_len].unpack('c*')
+    end
+
+    def mat
+      pt = super
+      pt[0, n * n].unpack('c*')
+    end
+
+    def read_len
+      readLen
+    end
+
+    def to_h
+      h = {}
+      %i[byte
+         word
+         read
+         mat
+         read_len
+         n
+         bias].each do |k|
+        h[k] = __send__(k)
+      end
+      h
+    end
+  end
+
+  class << self
+    def ssw_init(read, read_len, mat, n, score_size)
+      ptr = FFI.ssw_init(
+        read.pack('c*'), read_len, mat.flatten.pack('c*'), n, score_size
+      )
+      SSW::Profile.new(ptr)
+    end
+
+    def init_destroy(profile)
+      FFI.init_destroy(profile)
+    end
+
+    def ssw_align(prof, ref, ref_len, weight_gap0, weight_gapE, flag, filters, filterd, mask_len)
+      ptr = FFI.ssw_align(
+        prof, ref.pack('c*'), ref_len, weight_gap0, weight_gapE, flag, filters, filterd, mask_len
+      )
+      SSW::Align.new(ptr)
+    end
+
+    def align_destroy(align)
+      FFI.align_destroy(align)
+    end
+
+    def mark_mismatch(ref_begin1, read_begin1, read_end1, ref, read, read_len, cigar, cigar_len)
+      warn 'implementation: fiexme: **cigar' # FIXME
+      FFI.mark_mismatch(
+        ref_begin1, read_begin1, read_end1, ref.pack('c*'), read.pack('c*'), read_len, cigar, cigar_len.pack('l*')
+      )
+    end
+  end
 end

data/lib/libssw/ffi.rb

@@ -31,7 +31,7 @@ module LibSSW
       'uint16_t score2',
       'int32_t ref_begin1',
       'int32_t ref_end1',
-      'int32_t	read_begin1',
+      'int32_t read_begin1',
       'int32_t read_end1',
       'int32_t ref_end2',
       'uint32_t* cigar',
@@ -39,8 +39,8 @@ module LibSSW
     ]
 
     Profile = struct [
-      'int32_t* byte', # __m128i* profile_byte;	// 0: none
-      'int32_t* word', # __m128i* profile_word;	// 0: none
+      '__m128i* byte', # __m128i* profile_byte;	// 0: none
+      '__m128i* word', # __m128i* profile_word;	// 0: none
       'const int8_t* read',
       'const int8_t* mat',
       'int32_t readLen',
@@ -49,7 +49,12 @@ module LibSSW
     ]
 
     # s_profile* ssw_init (const int8_t* read, const int32_t readLen, const int8_t* mat, const int32_t n, const int8_t score_size)
-    try_extern 's_profile* ssw_init (const int8_t* read, int32_t readLen, const int8_t* mat, int32_t n, int8_t score_size)'
+    try_extern 's_profile* ssw_init ('  \
+               'const int8_t* read,'    \
+               'int32_t readLen,'       \
+               'const int8_t* mat,'     \
+               'int32_t n,'             \
+               'int8_t score_size)'
 
     try_extern 'void init_destroy (s_profile* p)'
 

data/lib/libssw/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module LibSSW
-  VERSION = '0.0.0.pre'
+  VERSION = '0.0.1.pre'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: libssw
 version: !ruby/object:Gem::Version
-  version: 0.0.0.pre
+  version: 0.0.1.pre
 platform: ruby
 authors:
 - kojix2
@@ -24,6 +24,20 @@ dependencies:
     - - ">="
       - !ruby/object:Gem::Version
         version: 1.0.7
+- !ruby/object:Gem::Dependency
+  name: bio
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
 - !ruby/object:Gem::Dependency
   name: bundler
   requirement: !ruby/object:Gem::Requirement