gene-matcher 0.1.4 → 0.1.6

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 3d31a639b4faa8f8c6b0625564377d8e43e2acd18e5aba9de696e04dd2c6de60
-  data.tar.gz: cea3e517c62db70b871dca04869622a3a5d89abaf618e0d3b36851a57ecb82fd
+  metadata.gz: 16d36396bec9b08dc0cd41f911274feef3b444c7c5408e406b528d627f74e92b
+  data.tar.gz: 71f75988056e3de2a0e022ed7b501e88209866f373e753e471298462ff7f428e
 SHA512:
-  metadata.gz: ad793ca6d4bb5e5b01bbf019e47d0835dbdbb02327683ffa68d82e10101bfb0b0b3c8202a8d0050b56cce0d9d209e03b5d045ec218b57e12fcc222f716697665
-  data.tar.gz: 87c261393441dbe6ac39ff701b2c73201beddddad184c6e0379a53aa9d25ee05609ec95a03ca7576b4c03aaf66ca0e282fedb34d5bb4a8478b1e615d9c91a128
+  metadata.gz: 47f79bf0bf8be63ada87103503d2ec0ef7717cf2d008ff836dbf9d5582c035f082e6049bb3add93c6130c15024a015bb7f52aa8f04300963cafed1119a0f1d4c
+  data.tar.gz: 38e05772aacdd96a57d43ff0cb16f19c00adf300c083d85b6fbb93cbb0449ab51212981d94ef1cbe5a38c542fa707d9bd9cb39dc0b58b3a3782ed9f5ab5eeda5

data/lib/alignment.rb

@@ -1,12 +1,10 @@
 class Alignment
-  attr_accessor :score, :alignmentI, :alignmentJ, :startI, :startJ, :endI, :endJ, :reversed, :aside, :source
+  attr_accessor :alignmentI, :alignmentJ, :startI, :startJ, :endI, :endJ, :reversed, :aside, :source
 
   BLANK = "-"
 
   def initialize(h=nil)
     return init_from_hash(h) if h
-    # スコア
-    @score = 0
     # 検索対象（データベースに入っていた）配列
     @alignmentI = ""
     # 検索した（クエリ=入力された）配列
@@ -27,7 +25,6 @@ class Alignment
 
   def init_from_hash(h)
     # 引数のハッシュから初期化
-    @score = h["score"]
     @alignmentI = h["alignmentI"]
     @alignmentJ = h["alignmentJ"]
     @startI = h["startI"]
@@ -39,6 +36,21 @@ class Alignment
     @source = h["source"]
   end
 
+  def to_h
+    # ハッシュに変換
+    h = {}
+    h["alignmentI"] = @alignmentI
+    h["alignmentJ"] = @alignmentJ
+    h["startI"] = @startI
+    h["startJ"] = @startJ
+    h["endI"] = @endI
+    h["endJ"] = @endJ
+    h["reversed"] = @reversed
+    h["aside"] = @aside
+    h["source"] = @source
+    return h
+  end
+
   # 二つの配列の一致部分と不一致部分を表した文字列を返す。
   # ex. AGTCAAAAAAAAA- :...:::::::::. AT-TAAAAAAAAAG
   # 
@@ -71,7 +83,17 @@ class Alignment
   def number_of_blank
   end
 
+  def score_a
+    return alignment_count / @alignmentI.length.to_f
+  end
+
+  def score(thresh=20)
+    # 長さがthresh以下なら0を返す
+    return 0 if @alignmentI.length < thresh
+    return alignment_count * score_a
+  end
+
   def to_s
-    "score=#{@score} I=" + @alignmentI + " J=" + @alignmentJ;
+    "score=#{score} I=" + @alignmentI + " J=" + @alignmentJ;
   end
 end

data/lib/gene-matcher.rb

@@ -14,8 +14,9 @@ class Matcher
   def scan(target_sequence,source = {})
     sw = SmithWaterman.instance
     a = sw.alignment(target_sequence, @input_sequence)
-    a.source = source
-    @alignments += [a] if a.score >= @limit
+    a.source = {target_sequence: target_sequence}.merge(source)
+    puts "#{a.score} / #{a.alignment_count} = #{a.score / a.alignment_count}" if ENV["DEBUG"]
+    @alignments += [a] if a.score / a.alignment_count >= @limit
   end
 end
 

data/lib/smith-waterman.rb

@@ -88,11 +88,7 @@ class SmithWaterman
     a.startI = i
     a.startJ = j
 
-    if a.alignmentI.length <= 20
-      a.score = 0
-    else
-      a.score = a.alignment_count ** 2 / a.alignmentI.length.to_f
-    end
+    # スコアはここでは求めない(Java版ではここで決定していた)
     return a
   end
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: gene-matcher
 version: !ruby/object:Gem::Version
-  version: 0.1.4
+  version: 0.1.6
 platform: ruby
 authors:
 - ITO Yosei
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2023-08-31 00:00:00.000000000 Z
+date: 2023-09-11 00:00:00.000000000 Z
 dependencies: []
 description: Algorithm for determining similar regions between nucleic acid sequences.
 email: y-itou@lumber-mill.co.jp