From 457568485ac3ef703ed93a0b1e7aa45a52affe3d Mon Sep 17 00:00:00 2001
From: depristo <depristo@348d0f76-0448-11de-a6fe-93d51630548a>
Date: Tue, 9 Feb 2010 01:21:04 +0000
Subject: [PATCH] simple Beagle input ROD

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2811 348d0f76-0448-11de-a6fe-93d51630548a
---
 .../sting/gatk/refdata/BeagleROD.java         | 87 +++++++++++++++++++
 1 file changed, 87 insertions(+)
 create mode 100755 java/src/org/broadinstitute/sting/gatk/refdata/BeagleROD.java

diff --git a/java/src/org/broadinstitute/sting/gatk/refdata/BeagleROD.java b/java/src/org/broadinstitute/sting/gatk/refdata/BeagleROD.java
new file mode 100755
index 000000000..60dcc1f36
--- /dev/null
+++ b/java/src/org/broadinstitute/sting/gatk/refdata/BeagleROD.java
@@ -0,0 +1,87 @@
+package org.broadinstitute.sting.gatk.refdata;
+
+import org.broadinstitute.sting.utils.GenomeLoc;
+import org.broadinstitute.sting.utils.GenomeLocParser;
+import org.broadinstitute.sting.utils.xReadLines;
+
+import java.util.*;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+import java.io.IOException;
+import java.io.File;
+import java.io.FileNotFoundException;
+
+public class BeagleROD extends BasicReferenceOrderedDatum {
+    GenomeLoc loc;
+    List<String> sampleNames = null;
+    Map<String, List<String>> sampleGenotypes = new HashMap<String, List<String>>();
+
+    public BeagleROD(String name) {
+        super(name);
+    }
+
+    public String toString() { return "BeagleRod"; }
+
+    public String delimiterRegex() {
+        return " ";
+    }
+
+    public GenomeLoc getLocation() {
+        return loc;
+    }
+
+    public List<String> getSampleNames() {
+        return sampleNames;
+    }
+
+    public Map<String, List<String>> getGenotypes() {
+        return sampleGenotypes;
+    }
+
+    public Object initialize(final File source) throws FileNotFoundException {
+        String firstLine = new xReadLines(source).next();
+        String[] parts = firstLine.split(" ");
+        if ( parts[0].equals("I") ) {
+            // I id NA12891 NA12891 NA12892 NA12892
+            sampleNames = Arrays.asList(parts).subList(2, parts.length);
+            return sampleNames;
+        } else {
+            throw new IllegalStateException("Beagle file " + source + " doesn't have required header line I");
+        }
+    }
+
+    private static Pattern MARKER_PATTERN = Pattern.compile("c(.+)_p([0-9]+)");
+
+    public static GenomeLoc parseMarkerName(String markerName) {
+        Matcher m = MARKER_PATTERN.matcher(markerName);
+        if ( m.matches() ) {
+            String contig = m.group(1);
+            long start = Long.valueOf(m.group(2));
+            return GenomeLocParser.createGenomeLoc(contig, start, start);
+        } else {
+            throw new IllegalArgumentException("Malformatted family structure string: " + markerName + " required format is mom+dad=child");
+        }
+    }
+
+    public boolean parseLine(final Object header, final String[] parts) throws IOException {
+        //System.out.printf("Parsing beagle parts=%s header=%s%n", parts, header);
+        List<String> sampleNames = (List<String>)header;
+
+        if ( parts.length == 0 || ! parts[0].equals("M") )
+            return false;
+        else {
+            loc = parseMarkerName(parts[1]);
+
+            for ( int i = 2; i < parts.length; i++ ) {
+                String sampleName = sampleNames.get(i-2);
+                if ( ! sampleGenotypes.containsKey(sampleName) ) {
+                    sampleGenotypes.put(sampleName, new ArrayList<String>());
+                }
+
+                sampleGenotypes.get(sampleName).add(parts[i]);
+            }
+
+            return true;
+        }
+    }
+}