parse_fasta 2.0.1 → 2.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 9f779807eb98f39dfbbd67376e04a890be3a7bf9
-  data.tar.gz: 5f7201962d03a7c3b2553d576d4721822dbc24a2
+  metadata.gz: 5b6a779e6af25bc031bbb8e4b0f0819723c5cb35
+  data.tar.gz: a919f21792fe97bdf9b8f0fe9064950f9fcaf19a
 SHA512:
-  metadata.gz: 9ddd0ae41bde830edba8af62be8ee88700e6c309b2f61ef0887a67579551d88308e0e53bbcb57c5d18ddd0cbc4b79f60a47986df14f2578415fa983ddbeb317f
-  data.tar.gz: ad89bb85f6cfcd8ed8af0d7c518d1b773cff2c9edcc5f51b241a149a55c217d070cbedca0946edb506b53ce430117d41ef7058a130e7e7e013dcc64c4d41b110
+  metadata.gz: 7fac0a0ca2c64235d5bf74221fd72373282e8f296045da23aec834350d6d2532d06847c1916b3ac22593c35a1e528239f5d015d3159d8054a3523af45fc6e299
+  data.tar.gz: e89410b0f7172737f811a68d5fc858b5cd0b60734b46715bb859c6a32cafbc95191b92fc49817807a079275df54d4a86a8538faca8386807c86429e559155672

data/.travis.yml

@@ -5,8 +5,7 @@ rvm:
   - 2.0
   - 2.1
   - 2.2
-  - 2.3
-
+  
 script: 'bundle exec rake'
 
 notifications:

data/CHANGELOG.md

@@ -1,5 +1,9 @@
 ## Versions ##
 
+### 2.1.0 ###
+
+Add `#to_s`, `#to_fasta`, and `#to_fastq` to `Record`.
+
 ### 2.0.0 ###
 
 A weird feature of `Zlib::GzipReader` made it so that if a gzipped file was created like this.

data/README.md

@@ -59,13 +59,29 @@ The `Record#desc` and `Record#qual` will be `nil` if the file you are parsing is
 ```ruby
 ParseFasta::SeqFile.open(ARGV[0]).each_record do |rec|
   if rec.qual
-    puts "@#{rec.header}"
-    puts rec.seq
-    puts "+#{rec.desc}"
-    puts rec.qual
+    # it's a fastQ record
   else
-    puts ">#{rec.header}"
-    puts rec.sequence
+    # it's a fastA record
   end  
 end
-```
+```
+
+You can also check this with `Record#fastq?`
+
+```ruby
+ParseFasta::SeqFile.open(ARGV[0]).each_record do |rec|
+  if rec.fastq?
+    # it's a fastQ record
+  else
+    # it's a fastA record
+  end  
+end
+```
+
+And there is a nice `#to_s` method, that does what it should whether the record is fastA or fastQ like. Check out the docs for info on the fancy `#to_fasta` and `#to_fastq` methods!
+
+```ruby
+ParseFasta::SeqFile.open(ARGV[0]).each_record do |rec|
+  puts rec.to_s
+end
+```

data/lib/parse_fasta/error.rb

@@ -24,6 +24,10 @@ module ParseFasta
     class ParseFastaError < StandardError
     end
 
+    # Raised when a method has a bad argument
+    class ArgumentError < ParseFastaError
+    end
+
     # Raised when the input file doesn't look like fastA or fastQ
     class DataFormatError < ParseFastaError
     end

data/lib/parse_fasta/record.rb

@@ -44,7 +44,7 @@ module ParseFasta
     # @param desc [String] the description line of a fastQ record
     # @param qual [String] the quality string of a fastQ record
     #
-    # @raise [SequenceFormatError] if a fastA sequence has a '>'
+    # @raise [ParseFasta::Error::SequenceFormatError] if a fastA sequence has a '>'
     #   character in it
     def initialize args = {}
       @header = args.fetch :header
@@ -56,7 +56,7 @@ module ParseFasta
 
       seq = args.fetch(:seq).gsub(/\s+/, "")
 
-      if @qual # is fastQ
+      if fastq? # is fastQ
         @seq = seq
       else # is fastA
         @seq = check_fasta_seq(seq)
@@ -73,6 +73,93 @@ module ParseFasta
           self.desc == rec.desc && self.qual == rec.qual
     end
 
+    # Return a fastA or fastQ record ready to print.
+    #
+    # If the Record is fastQ like then it returns a fastQ record
+    # string. If the record is fastA like, then it returns a fastA
+    # record string.
+    #
+    # @return [String] a printable sequence record
+    #
+    # @example When the record is fastA like
+    #   rec = Record.new header: "Apple", seq: "ACTG"
+    #   rec.to_s #=> ">Apple\nACTG"
+    #
+    # @example When the record is fastQ like
+    #   rec = Record.new header: "Apple", seq: "ACTG", desc: "Hi", qual: "IIII"
+    #   rec.to_s #=> "@Apple\nACTG\n+Hi\nIIII"
+    def to_s
+      if fastq?
+        to_fastq
+      else
+        to_fasta
+      end
+    end
+
+    # Returns a fastA record ready to print.
+    #
+    # If the record is fastQ like, the desc and qual are dropped.
+    #
+    # @return [String] a printable fastA sequence record
+    #
+    # @example When the record is fastA like
+    #   rec = Record.new header: "Apple", seq: "ACTG"
+    #   rec.to_fasta #=> ">Apple\nACTG"
+    #
+    # @example When the record is fastQ like
+    #   rec = Record.new header: "Apple", seq: "ACTG", desc: "Hi", qual: "IIII"
+    #   rec.to_fasta #=> ">Apple\nACTG"
+    def to_fasta
+      ">#{header}\n#{seq}"
+    end
+
+    # Returns a fastA record ready to print.
+    #
+    # If the record is fastA like, the desc and qual can be specified.
+    #
+    # @return [String] a printable fastQ sequence record
+    #
+    # @example When the record is fastA like, no args
+    #   rec = Record.new header: "Apple", seq: "ACTG"
+    #   rec.to_fastq #=> "@Apple\nACTG\n+\nIIII"
+    #
+    # @example When the record is fastA like, desc and qual specified
+    #   rec = Record.new header: "Apple", seq: "ACTG"
+    #   rec.to_fastq decs: "Hi", qual: "A" #=> "@Apple\nACTG\n+Hi\nAAAA"
+    #
+    # @example When the record is fastA like, can specify fancy qual strings
+    #   rec = Record.new header: "Apple", seq: "ACTGACTG"
+    #   rec.to_fastq decs: "Hi", qual: "!a2" #=> "@Apple\nACTG\n+Hi\n!a2!a2!a"
+    #
+    # @example When the record is fastQ like
+    #   rec = Record.new header: "Apple", seq: "ACTG", desc: "Hi", qual: "IIII"
+    #   rec.to_fastq #=> ">Apple\nACTG"
+    #
+    # @raise [ParseFasta::Error::ArgumentError] if qual is ""
+    def to_fastq opts = {}
+      if fastq?
+        "@#{@header}\n#{@seq}\n+#{@desc}\n#{qual}"
+      else
+        qual = opts.fetch :qual, "I"
+        check_qual qual
+
+        desc  = opts.fetch :desc, ""
+
+        qual_str = make_qual_str qual
+
+        "@#{@header}\n#{@seq}\n+#{desc}\n#{qual_str}"
+      end
+    end
+
+    # Returns true if record is a fastQ record.
+    #
+    # This method returns true if the fastq instance method is set.
+    #
+    # @return [Bool] true if record is fastQ, false if it is fastA
+    def fastq?
+      true if @qual
+    end
+
     private
 
     def check_fasta_seq seq
@@ -84,5 +171,16 @@ module ParseFasta
         seq
       end
     end
+
+    def make_qual_str qual
+      (qual * (@seq.length / qual.length.to_f).ceil)[0, @seq.length]
+    end
+
+    def check_qual qual
+      if qual.length.zero?
+        raise ParseFasta::Error::ArgumentError,
+              ":qual was '#{qual.inspect}', but it can't be empty"
+      end
+    end
   end
 end

data/lib/parse_fasta/version.rb

@@ -17,5 +17,5 @@
 # along with parse_fasta.  If not, see <http://www.gnu.org/licenses/>.
 
 module ParseFasta
-  VERSION = "2.0.1"
+  VERSION = "2.1.0"
 end

data/spec/parse_fasta/record_spec.rb

@@ -109,7 +109,120 @@ module ParseFasta
 
         expect(rec1 == rec2).to eq false
       end
+    end
+
+    describe "#fastq?" do
+      it "returns true if @qual is not nil" do
+        rec = Record.new header: "a", seq: "a", qual: "a"
+
+        expect(rec.fastq?).to eq true
+      end
+
+      it "returns nil if @qual is nil" do
+        rec = Record.new header: "a", seq: "a"
+
+        expect(rec.fastq?).to eq nil
+      end
+    end
+
+    describe "#to_s" do
+      context "when the record is fastA like" do
+        it "returns a string of the fastA record ready to print" do
+          rec = Record.new header: "apple", seq: "actg"
+
+          expect(rec.to_s).to eq ">apple\nactg"
+        end
+      end
+
+      context "when the record is fastQ like" do
+        it "returns a string of the fastQ record ready to print" do
+          rec = Record.new header: "apple", seq: "actg", desc: "", qual: "IIII"
+
+          expect(rec.to_s).to eq "@apple\nactg\n+\nIIII"
+        end
+      end
+    end
+
+    describe "#to_fasta" do
+      context "when the record is fastA like" do
+        it "returns a string of the fastA record ready to print" do
+          rec = Record.new header: "apple", seq: "actg"
+
+          expect(rec.to_fasta).to eq ">apple\nactg"
+        end
+      end
+
+      context "when the record is fastQ like" do
+        it "returns a string of the fastQ record in fastA format" do
+          rec = Record.new header: "apple", seq: "actg", desc: "", qual: "IIII"
+
+          expect(rec.to_fasta).to eq ">apple\nactg"
+        end
+      end
+    end
 
+    describe "#to_fastq" do
+      context "when the record is fastA like" do
+        let(:rec) {Record.new header: "apple", seq: "actg"}
+
+        it "has a default quality string (I) and description" do
+          expect(rec.to_fastq).to eq "@apple\nactg\n+\nIIII"
+        end
+
+        context "specifying values for qual string" do
+          it "can specify the qual string value (single value)" do
+            expect(rec.to_fastq qual: "A").to eq "@apple\nactg\n+\nAAAA"
+          end
+
+          it "can specify complex qual string values (uneven division)" do
+            seq  = "actgactgactgactg"
+            qual = "1@c1@c1@c1@c1@c1"
+            rec = Record.new header: "apple", seq: seq
+
+            expect(rec.to_fastq qual: "1@c").to eq "@apple\n#{seq}\n+\n#{qual}"
+          end
+
+          it "can specify complex qual string values (even division)" do
+            seq  = "actgactgactgactg"
+            qual = "a1a1a1a1a1a1a1a1"
+            rec = Record.new header: "apple", seq: seq
+
+            expect(rec.to_fastq qual: "a1").to eq "@apple\n#{seq}\n+\n#{qual}"
+          end
+
+          it "can specify complex qual string values (full length)" do
+            seq  = "actgactgactgactg"
+            qual = "a1a1a1a1a1a1a1a1"
+            rec = Record.new header: "apple", seq: seq
+
+            expect(rec.to_fastq qual: qual).to eq "@apple\n#{seq}\n+\n#{qual}"
+          end
+
+          it "raises ParseFasta::Error::ArgumentError when :qual is empty" do
+            seq  = "actgactgactgactg"
+            qual = "IIIIIIIIIIIIIIII"
+            rec = Record.new header: "apple", seq: seq
+
+            expect{rec.to_fastq qual: ""}.to raise_error ParseFasta::Error::ArgumentError
+          end
+        end
+
+        it "can specify the description" do
+          expect(rec.to_fastq desc: "pie").to eq "@apple\nactg\n+pie\nIIII"
+        end
+
+        it "can specify the both" do
+          expect(rec.to_fastq qual: "A", desc: "pie").to eq "@apple\nactg\n+pie\nAAAA"
+        end
+      end
+
+      context "when the record is fastQ like" do
+        it "returns a string of the fastQ format" do
+          rec = Record.new header: "apple", seq: "actg", desc: "", qual: "IIII"
+
+          expect(rec.to_fastq).to eq "@apple\nactg\n+\nIIII"
+        end
+      end
     end
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: parse_fasta
 version: !ruby/object:Gem::Version
-  version: 2.0.1
+  version: 2.1.0
 platform: ruby
 authors:
 - Ryan Moore
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-10-21 00:00:00.000000000 Z
+date: 2016-10-22 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler