From: Jim Procter <jprocter@issues.jalview.org>
Date: Mon, 8 Jun 2020 10:08:25 +0000 (+0100)
Subject: JAL-3641 prototype FASTQ import via groovy script for 2.11.1
X-Git-Url: http://source.jalview.org/gitweb/?a=commitdiff_plain;h=49492cdc61a03e377534696dc78ae76aae189adb;p=jalview.git

JAL-3641 prototype FASTQ import via groovy script for 2.11.1
---

diff --git a/examples/groovy/FastQParser.groovy b/examples/groovy/FastQParser.groovy
new file mode 100644
index 0000000..7658539
--- /dev/null
+++ b/examples/groovy/FastQParser.groovy
@@ -0,0 +1,155 @@
+import htsjdk.samtools.cram.encoding.readfeatures.BaseQualityScore
+import htsjdk.samtools.fastq.FastqReader
+import htsjdk.samtools.fastq.FastqRecord
+import jalview.datamodel.AlignmentI
+import jalview.datamodel.SequenceI
+import jalview.datamodel.Sequence
+import jalview.datamodel.SequenceFeature
+import jalview.io.FileFormatI
+import jalview.io.FileFormats
+import jalview.io.FileParse
+import jalview.io.AlignFile
+import jalview.io.DataSourceType
+
+/*
+ * Prototype Jalview FASTQ importer based on HTS JDK 2.20
+ */
+
+/*
+ * EDIT THE FILE PATH BELOW TO IMPORT A FASTQ FILE
+ */
+FASTQ_FILE = "/Users/jprocter/Downloads/small_normalised_illumina_adeno.fastq.gz"
+FASTQ_FILE = "/Users/jprocter/Downloads/test.fastq"
+
+
+/*
+ * A parser class to read or write the format
+ * based on the fileFormat.groovy example for creating your own alignment parser/writer
+ */
+class FastQFile extends AlignFile 
+{
+  FastqReader fqreader;
+  @Override
+  protected boolean isParseImmediately()
+  {
+    return false;
+  }
+  /*
+   * Constructor for reading a file; the superclass
+   * constructor will call the parse() method
+   */
+  FastQFile(FileParse src) 
+  {
+    super(src)
+  }
+
+  /*
+   * Constructor for writing out an alignment
+   */
+  FastQFile(AlignmentI al) 
+  {
+    throw new Error("Unimplemented");
+  }
+  
+  /*
+   * Parse a formatted data file (with no error checking!)
+   */
+  void parse() 
+  {
+    fqreader = new FastqReader(getReader());
+
+    FastqRecord record
+    while (fqreader.hasNext() && (record = fqreader.next())!=null)
+    {
+      // Create a sequence from read name and sequence string. start & end are 1-length
+      Sequence seq = new Sequence(record.getReadName(), record.getReadString());
+      byte[] qualities = record.getBaseQualities();
+      if (qualities!=null)
+      {
+        
+        // Two options here: create an annotation row, or create features.
+        // annotations take far longer to create and manipulate in Jalview 2.11 (this is changing in 2.12!)
+        // so we create sequence features for now
+        boolean makeAnn=false;
+        
+        jalview.datamodel.Annotation[] anns=new jalview.datamodel.Annotation[qualities.length];
+        int p = 0;
+        for (byte q:qualities)
+        {
+          if (makeAnn) { anns[p++]=new jalview.datamodel.Annotation(Float.valueOf(q)) }
+          else { 
+            SequenceFeature sf = new SequenceFeature("Phred","Score", p+1,p+1,Float.valueOf(q),"");
+            seq.addSequenceFeature(sf);
+            p++;
+          }
+        }
+        if (makeAnn) {
+          jalview.datamodel.AlignmentAnnotation qalAnnot = new jalview.datamodel.AlignmentAnnotation("PhredScore","Base qualities: "+record.getBaseQualityHeader(),anns);
+          qalAnnot.visible=false; // hidden by default
+          seq.addAlignmentAnnotation(qalAnnot);
+          annotations.add(qalAnnot);
+        }
+      }
+      addSequence(seq)
+    }
+    fqreader.close();
+  }
+  
+  /*
+   * Print the formatted sequences
+   * (addSuffix always defaults to true as no user preference for it)
+   */
+  String print(SequenceI[] seqs, boolean addSuffix) 
+  {
+      // not supported
+  }
+}
+
+/*
+ * Define and register the fileformat class for handling FastQ
+ */
+def myFormat = { ->
+  [
+    getName: { -> 'FASTQ Parser (Groovy)' },
+    
+    toString: { -> getName() },
+    
+    getExtensions: { 'fq,fastq'},
+    
+    getReader: { FileParse source -> new FastQFile(source) },
+    
+    getWriter: { AlignmentI al -> new FastQFile(al) },
+    
+    isReadable: { -> true },
+    
+    isWritable: { -> false },
+    
+    isTextFormat: { -> true },
+    
+    isStructureFile: { -> false },
+    
+    isComplexAlignFile: { -> false },
+
+   ] as FileFormatI
+}
+
+// read in the file specified at the top of the Groovy script
+
+FastQFile parser = new FastQFile(new FileParse(FASTQ_FILE, DataSourceType.FILE));
+AlignmentI al = new jalview.datamodel.Alignment(parser.getSeqsAsArray());
+parser.addAnnotations(al);
+jalview.gui.AlignFrame newwindow = new jalview.gui.AlignFrame(al,800,600);
+jalview.gui.Desktop.addInternalFrame(newwindow, parser.getDataName(), 800,600);
+
+
+/*
+ * Register the file format. After running this script in Jalview's
+ * Groovy console, the new format should be shown in open file.
+ *
+ * However, we don't yet have a way of dynamically updating the IdentifyFile routine,
+ * so a FastQ file selected in the file browser won't be imported using this reader
+ *
+ */
+
+ // FileFormats.instance.registerFileFormat(myFormat())
+