Merge branch 'master' of https://github.com/dotnetbio/bio

src/Source/Framework/Bio.Core/IO/GenBank/GenBankParser.cs

@@ -1042,7 +1042,11 @@ namespace Bio.IO.GenBank
             {
                 // Using a regex is too slow.
                 int len = line.Length;
-                int k = line.Length > 8 && line[8] == ' ' ? 9 : 10;
+
+                int k = 0;
+                while (k < len && (line[k] == ' ' || Char.IsNumber(line[k])))
+                    k++;
+
                 while (k < len)
                 {
                     string seqData = line.Substring(k, Math.Min(10, len - k));

src/Source/Tests/Bio.Tests/IO/GenBank/GenBankTests.cs

@@ -32,6 +32,7 @@ namespace Bio.Tests.IO.GenBank
         #region Fields
         private const string _genBankFile_LocusTokenParserTest = @"TestUtils\GenBank\LocusTokenParserTest.gb";
         private const string _genBankFile_ParseOriginShifted = @"TestUtils\GenBank\ParseOriginShifted.gb";
+        private const string _genBankFile_ParseOriginShifted2 = @"TestUtils\GenBank\ParseOriginShifted2.gb";
         private const string _genBankFile_ParseVersionEmpty = @"TestUtils\GenBank\ParseVersionEmptyTest.gb";
         private const string _genBankFile_EmptyOrganismClassificationTest = @"TestUtils\GenBank\EmptyOrganismClassificationTest.gb";
         private const string _genBankFile_WithTPAPrimaryData = @"TestUtils\GenBank\BK000016-tpa.gbk";
@@ -313,6 +314,20 @@ ORIGIN
             Assert.IsNotNull(seq);
         }
 
+        /// <summary>
+        /// This test used to fail.  The sequence is in a GenBank format and would fail if the
+        /// ORIGIN contents started at any other index than 10.
+        /// </summary>
+        [Test]
+        [Category("Priority0")]
+        public void TestGenBankParseOriginShifted2()
+        {
+            // parse
+            GenBankParser parser = new GenBankParser();
+            ISequence seq = parser.Parse(_genBankFile_ParseOriginShifted2).FirstOrDefault();
+            Assert.IsNotNull(seq);
+        }
+
         /// <summary>
         /// Verifies that the parser throws an exception when calling ParseOne on a file containing
         /// more than one sequence.

src/Source/Tests/TestData/TestData.projitems

@@ -1091,6 +1091,9 @@
     <None Include="$(MSBuildThisFileDirectory)TestUtils\GenBank\NM_001747.gb">
       <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
     </None>
+    <None Include="$(MSBuildThisFileDirectory)TestUtils\GenBank\ParseOriginShifted2.gb">
+      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+    </None>
     <None Include="$(MSBuildThisFileDirectory)TestUtils\GenBank\ParseOriginShifted.gb">
       <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
     </None>
@@ -2184,7 +2187,7 @@
       <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
     </None>
     <None Include="$(MSBuildThisFileDirectory)TestUtils\PacBio\ccs.bam">
-          <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
     </None>
   </ItemGroup>
   <ItemGroup>

src/Source/Tests/TestData/TestUtils/GenBank/ParseOriginShifted2.gb

@@ -0,0 +1,32 @@
+LOCUS       AAB24880     74 aa            linear   DNA 08-MAY-1993
+DEFINITION  zinc finger [Homo sapiens].
+ACCESSION   AAB24880
+VERSION     .
+DBSOURCE    accession S52506.1
+KEYWORDS    .
+SOURCE      Homo sapiens (human)
+  ORGANISM  Homo sapiens
+            .
+REFERENCE   1  (residues 1 to 74)
+  AUTHORS   Saleh,M., Selleri,L., Little,P.F. and Evans,G.A.
+  TITLE     Isolation and expression of linked zinc finger gene clusters on
+            human chromosome 11q
+  JOURNAL   Genomics 14 (4), 970-978 (1992)
+   PUBMED   1339395
+  REMARK    GenBank staff at the National Library of Medicine created this
+            entry [NCBI gibbsq 122386] from the original journal article.
+COMMENT     Method: conceptual translation supplied by author.
+FEATURES             Location/Qualifiers
+     source          1..74
+                     /organism="Homo sapiens"
+                     /db_xref="taxon:9606"
+     Protein         1..74
+                     /name="zinc finger"
+     CDS             1..74
+                     /gene="ZNF123"
+                     /coded_by="S52506.1:1..222"
+ORIGIN      
+      1 CCTTCCATCG ATCGGAAAGA CTAGCTAGGG GCTGCAGGAA TTCGGGACAA AAGTCGCTTC 
+     61 GAAGCGGCGC GCCCCCGGAC CGGCTGTACA GGATTTAAAT GCGCTAGCGA ATTCTCATCA 
+      
+//