diff --git a/uk/ac/sanger/artemis/components/variant/IOUtils.java b/uk/ac/sanger/artemis/components/variant/IOUtils.java
index 407dab698908b897716c3883a461a71ebe8c5b28..80ec4dff8547a9ccee3087a6293c9b9bfc88332e 100644
--- a/uk/ac/sanger/artemis/components/variant/IOUtils.java
+++ b/uk/ac/sanger/artemis/components/variant/IOUtils.java
@@ -23,13 +23,18 @@
*/
package uk.ac.sanger.artemis.components.variant;
+import java.io.BufferedReader;
import java.io.File;
import java.io.FileInputStream;
+import java.io.FileOutputStream;
+import java.io.FileReader;
import java.io.FileWriter;
import java.io.IOException;
import java.io.InputStream;
+import java.io.PrintWriter;
import java.io.Writer;
import java.net.URL;
+import java.util.Hashtable;
import java.util.List;
import java.util.Map;
import java.util.Vector;
@@ -167,6 +172,276 @@ class IOUtils
new MessageDialog (null, "Saved Files", filterFiles, false);
}
+ /**
+ * Export all variant sites to a multiple fasta file.
+ * @param vcfView
+ */
+ protected static void exportVariantFasta(final VCFview vcfView)
+ {
+ final EntryGroup entryGroup = vcfView.getEntryGroup();
+ final int length = entryGroup.getSequenceLength();
+ final Bases bases = entryGroup.getBases();
+ FeatureVector features = entryGroup.getAllFeatures();
+
+ final String name = entryGroup.getActiveEntries().elementAt(0).getName();
+ final File newfile = new File(
+ getBaseDirectoryFromEntry(entryGroup.getActiveEntries().elementAt(0)),
+ name);
+ try
+ {
+ final File f = getFile(newfile.getAbsolutePath(), 1, ".fasta", null);
+ if(f == null)
+ return;
+
+ int ntotalSamples = 0;
+ for (int i = 0; i < vcfView.getVcfReaders().length; i++)
+ ntotalSamples += vcfView.getVcfReaders()[i].getNumberOfSamples();
+
+ final Writer[] writer = new Writer[ntotalSamples];
+ final File[] tmpFiles = new File[ntotalSamples];
+
+ for (int i = 0; i < vcfView.getVcfReaders().length; i++)
+ {
+ final String names[] = vcfView.getVcfReaders()[i].sampleNames;
+ for(int j=0; j<names.length; j++)
+ {
+ tmpFiles[i] = File.createTempFile(names[j].replaceAll("[/\\:]", "_"), "art");
+ writer[i] = new FileWriter( tmpFiles[i] );
+ writer[i].write(">"+names[j]);
+ }
+ }
+
+ int MAX_BASE_CHUNK = (10000/ntotalSamples)*SEQUENCE_LINE_BASE_COUNT;
+ //System.out.println("MAX_BASE_CHUNK "+MAX_BASE_CHUNK);
+
+ final Hashtable<Integer, VCFRecord> records[] = new Hashtable[MAX_BASE_CHUNK];
+ int baseCount = 0;
+
+ // write variant sites to temp files
+ for(int i=0; i<length; i+=MAX_BASE_CHUNK)
+ {
+ int end = i+MAX_BASE_CHUNK;
+ int start = i+1;
+ if(end > length)
+ end = length;
+
+ storeVCFRecords(vcfView, records, start, end);
+ baseCount = writeVariants(vcfView, records, writer, features, ntotalSamples, start, end, baseCount);
+ }
+
+ for(int i=0; i<ntotalSamples; i++)
+ writer[i].close();
+
+ // concatenate the single fasta files into a multiple fasta file
+ final PrintWriter pw = new PrintWriter(new FileOutputStream(f));
+ for (int i = 0; i < tmpFiles.length; i++)
+ {
+ final BufferedReader br = new BufferedReader(new FileReader(tmpFiles[i].getPath()));
+ String line;
+ while ( (line = br.readLine()) != null)
+ pw.println(line);
+ br.close();
+ tmpFiles[i].delete();
+ }
+ pw.close();
+ }
+ catch(IOException e)
+ {
+ e.printStackTrace();
+ }
+ }
+
+ /**
+ * For a given range store the VFFRecord in a Hashtable.
+ * @param vcfView
+ * @param records
+ * @param start
+ * @param end
+ * @throws IOException
+ */
+ private static void storeVCFRecords(final VCFview vcfView,
+ final Hashtable<Integer, VCFRecord> records[],
+ final int start,
+ final int end) throws IOException
+ {
+
+ for (int i = 0; i < vcfView.getVcfReaders().length; i++)
+ records[i] = new Hashtable<Integer, VCFRecord> ();
+
+ for (int i = 0; i < vcfView.getVcfReaders().length; i++)
+ {
+ AbstractVCFReader reader = vcfView.getVcfReaders()[i];
+ if(vcfView.isConcatenate())
+ {
+ String[] contigs = reader.getSeqNames();
+ for(int j=0; j<contigs.length; j++)
+ {
+ int offset = vcfView.getSequenceOffset(contigs[j]);
+ int nextOffset;
+ if(j<contigs.length-1)
+ nextOffset = vcfView.getSequenceOffset(contigs[j+1]);
+ else
+ nextOffset = vcfView.seqLength;
+
+ if( (start >= offset && start <= nextOffset) ||
+ (end >= offset && end <= nextOffset) )
+ {
+ int thisStart = start - offset;
+ if(thisStart < 1)
+ thisStart = 1;
+ loadRecords(records[i], reader, contigs[j], thisStart, end - offset, offset);
+ }
+ }
+ }
+ else
+ loadRecords(records[i], reader, vcfView.getChr(), start, end, 0);
+ }
+ }
+
+ private static void loadRecords(Hashtable<Integer, VCFRecord> records,
+ final AbstractVCFReader reader,
+ final String contig,
+ final int start,
+ final int end,
+ final int offset) throws IOException
+ {
+ VCFRecord record;
+ while((record = reader.getNextRecord(contig, start, end)) != null)
+ {
+ if(records == null)
+ records = new Hashtable<Integer, VCFRecord> ();
+ records.put(record.getPos()+offset, record);
+ }
+ }
+
+ private static int writeVariants(final VCFview vcfView,
+ final Hashtable<Integer, VCFRecord> records[],
+ final Writer writer[],
+ final FeatureVector features,
+ final int ntotalSamples,
+ final int start,
+ final int end,
+ int bc) throws IOException
+ {
+ for (int i = start; i < end; i++)
+ {
+ int basePosition = i;
+ int thisSample = 0;
+ final String[] thisBase = new String[ntotalSamples];
+ boolean seenSNP = false;
+ int insertionLength = 0;
+
+ // loop over each VCF file
+ for (int j = 0; j < vcfView.getVcfReaders().length; j++)
+ {
+ AbstractVCFReader reader = vcfView.getVcfReaders()[j];
+ VCFRecord record = records[j].get(i);
+
+ if(record == null)
+ continue;
+
+ boolean vcf_v4 = reader.isVcf_v4();
+ int nsamples = reader.getNumberOfSamples();
+ // loop over each sample
+ for(int k=0; k<nsamples; k++)
+ {
+ // look at each type of variant
+ if(vcfView.showVariant(record, features, basePosition, reader, k, j) )
+ {
+ if(record.getAlt().isDeletion(vcf_v4))
+ {
+ thisBase[thisSample] = "-";
+ }
+ else if(record.getAlt().isInsertion(vcf_v4))
+ {
+ String in = record.getAlt().toString();
+ if(in.startsWith("I"))
+ in = in.substring(1);
+ thisBase[thisSample] = in;
+ if(in.length() > insertionLength)
+ insertionLength = in.length();
+ }
+ else if(record.getAlt().isMultiAllele(k))
+ {
+ String base = MultipleAlleleVariant.getIUBCode(record);
+ if(base != null)
+ {
+ thisBase[thisSample] = base;
+ seenSNP = true;
+ }
+ }
+ else if(record.getAlt().isNonVariant())
+ thisBase[thisSample] = ".";
+ else
+ {
+ thisBase[thisSample] = record.getAlt().toString();
+ seenSNP = true;
+ }
+ }
+ else
+ thisBase[thisSample] = "N"; // filtered out
+
+ thisSample++;
+ }
+ }
+
+ if(seenSNP)
+ {
+ int remainder = 0;
+ for(int j=0; j<thisBase.length; j++)
+ {
+ if(thisBase[j] != null)
+ {
+ for(int k=0; k<thisBase[j].length(); k++)
+ {
+ remainder = (bc+k)%SEQUENCE_LINE_BASE_COUNT;
+
+ /*if(j==0 && k==0 && bc>3299 && bc<3365)
+ System.out.println("HERE "+thisBase[j]+" "+bc+" "+basePosition+" "+remainder);*/
+
+ if(remainder == 0)
+ writer[j].write(System.getProperty("line.separator"));
+ writer[j].write(thisBase[j].charAt(k));
+ }
+ }
+ else
+ {
+ remainder = bc%SEQUENCE_LINE_BASE_COUNT;
+ if(remainder == 0)
+ writer[j].write(System.getProperty("line.separator"));
+ writer[j].write("N");
+ }
+
+ if(insertionLength > 0)
+ {
+ int ins;
+ if(thisBase[j] != null)
+ ins = insertionLength-thisBase[j].length();
+ else
+ ins = insertionLength-1;
+
+ int rem = remainder+1;
+ for(int k=0; k<ins; k++)
+ {
+ remainder = (rem+k)%SEQUENCE_LINE_BASE_COUNT;
+ if(remainder == 0)
+ writer[j].write(System.getProperty("line.separator"));
+ writer[j].write("-");
+ }
+ }
+ }
+
+ if(insertionLength > 0)
+ bc+=insertionLength;
+ else
+ bc++;
+ }
+ }
+
+ return bc;
+ }
+
+
/**
* Write out FASTA for a selected base range
* @param vcfView
@@ -406,7 +681,8 @@ class IOUtils
header.append(reader.getName()).append(" ");
header.append(seqName).append(" ");
header.append(sbeg).append(":").append(send);
- header.append((marker.isForwardMarker() ? "" : " reverse"));
+ if(marker != null)
+ header.append((marker.isForwardMarker() ? "" : " reverse"));
return header;
}
diff --git a/uk/ac/sanger/artemis/components/variant/VCFview.java b/uk/ac/sanger/artemis/components/variant/VCFview.java
index 00a1f2bbe22a9c982194a98ff370d081f60e3f9f..6fb8b821c7dce39c1c7af354bec766aa8ba3ac05 100644
--- a/uk/ac/sanger/artemis/components/variant/VCFview.java
+++ b/uk/ac/sanger/artemis/components/variant/VCFview.java
@@ -466,6 +466,7 @@ public class VCFview extends JPanel
}
});
export.add(exportVCF);
+
export.add(new JSeparator());
final JMenuItem exportFastaSelected = new JMenuItem("FASTA of selected feature(s) ...");
@@ -504,6 +505,26 @@ public class VCFview extends JPanel
});
export.add(exportFasta);
+ final JMenuItem viewMinimalFasta = new JMenuItem("FASTA of variant sites only ...");
+ viewMinimalFasta.addActionListener(new ActionListener(){
+ public void actionPerformed(ActionEvent e)
+ {
+ Container f = getVcfContainer();
+ try
+ {
+ f.setCursor(new Cursor(Cursor.WAIT_CURSOR));
+ IOUtils.exportVariantFasta(VCFview.this);
+ }
+ finally
+ {
+ f.setCursor(new Cursor(Cursor.DEFAULT_CURSOR));
+ }
+ }
+ });
+ export.addSeparator();
+ export.add(viewMinimalFasta);
+
+
final JMenu view = new JMenu("View");
popup.add(view);
final JMenuItem viewFastaSelected = new JMenuItem("FASTA of selected feature(s) ...");
@@ -541,7 +562,8 @@ public class VCFview extends JPanel
}
});
view.add(viewFasta);
-
+
+
JMenu graph = new JMenu("Graph");
popup.add(graph);