From 125e6c1a475a25388cd2ad32a3f2397772673d65 Mon Sep 17 00:00:00 2001
From: Mauricio Carneiro <carneiro@broadinstitute.org>
Date: Fri, 6 Jul 2012 15:03:20 -0400
Subject: [PATCH] added BinaryTagCovariate for ancient dna analysis

---
 .../gatk/walkers/bqsr/BinaryTagCovariate.java | 61 +++++++++++++++++++
 .../sting/gatk/walkers/bqsr/Covariate.java    |  2 +-
 .../bqsr/RecalibrationArgumentCollection.java |  7 +++
 .../walkers/bqsr/RecalibrationReport.java     |  3 +
 4 files changed, 72 insertions(+), 1 deletion(-)
 create mode 100644 public/java/src/org/broadinstitute/sting/gatk/walkers/bqsr/BinaryTagCovariate.java

diff --git a/public/java/src/org/broadinstitute/sting/gatk/walkers/bqsr/BinaryTagCovariate.java b/public/java/src/org/broadinstitute/sting/gatk/walkers/bqsr/BinaryTagCovariate.java
new file mode 100644
index 000000000..424d076b7
--- /dev/null
+++ b/public/java/src/org/broadinstitute/sting/gatk/walkers/bqsr/BinaryTagCovariate.java
@@ -0,0 +1,61 @@
+package org.broadinstitute.sting.gatk.walkers.bqsr;
+
+import org.broadinstitute.sting.utils.exceptions.UserException;
+import org.broadinstitute.sting.utils.sam.GATKSAMRecord;
+
+/**
+ * Binary covariate allows BQSR to recalibrate based on a binary covariate in the BAM file. This covariate should assume values of 1 and 0.
+ *
+ * @author Mauricio Carneiro
+ * @since 7/6/12
+ */
+public class BinaryTagCovariate implements StandardCovariate {
+
+    private String tag;
+
+    @Override
+    public void initialize(RecalibrationArgumentCollection RAC) {
+        tag = RAC.BINARY_TAG_NAME;
+    }
+
+    @Override
+    public void recordValues(GATKSAMRecord read, ReadCovariates values) {
+        final Object tagObject = read.getAttribute(tag);
+
+        byte[] binaryTag;
+        if (tagObject instanceof byte[])
+            binaryTag = (byte[]) tagObject;
+        else if (tagObject instanceof String) {
+            int readLength = ((String) tagObject).length();
+            binaryTag = new byte[readLength];
+            for (int i = 0; i<readLength; i++)
+                binaryTag[i] = Byte.decode(((String) tagObject).substring(i, i+1));
+        }
+        else
+            throw new UserException("Binary tag is not a byte array (fast) or a string (slow). Type not supported");
+
+        for (int i = 0; i < read.getReadLength(); i++) {
+            values.addCovariate((int) binaryTag[i], (int) binaryTag[i], (int) binaryTag[i], i);
+        }
+    }
+
+    @Override
+    public Object getValue(String str) {
+        return Integer.decode(str);
+    }
+
+    @Override
+    public String formatKey(int key) {
+        return String.format("%d", key);
+    }
+
+    @Override
+    public int keyFromValue(Object value) {
+        return Integer.decode((String) value);
+    }
+
+    @Override
+    public int maximumKeyValue() {
+        return 1;
+    }
+}
diff --git a/public/java/src/org/broadinstitute/sting/gatk/walkers/bqsr/Covariate.java b/public/java/src/org/broadinstitute/sting/gatk/walkers/bqsr/Covariate.java
index 4d22eb409..1ad5346fa 100755
--- a/public/java/src/org/broadinstitute/sting/gatk/walkers/bqsr/Covariate.java
+++ b/public/java/src/org/broadinstitute/sting/gatk/walkers/bqsr/Covariate.java
@@ -55,7 +55,7 @@ public interface Covariate {
     public void recordValues(final GATKSAMRecord read, final ReadCovariates values);
 
     /**
-     * Used to get the covariate's value from input csv file during on-the-fly recalibration
+     * Used to get the covariate's value from input (Recalibration Report) file during on-the-fly recalibration
      *
      * @param str the key in string type (read from the csv)
      * @return the key in it's correct type.
diff --git a/public/java/src/org/broadinstitute/sting/gatk/walkers/bqsr/RecalibrationArgumentCollection.java b/public/java/src/org/broadinstitute/sting/gatk/walkers/bqsr/RecalibrationArgumentCollection.java
index 2688fe655..9d7ec7b26 100755
--- a/public/java/src/org/broadinstitute/sting/gatk/walkers/bqsr/RecalibrationArgumentCollection.java
+++ b/public/java/src/org/broadinstitute/sting/gatk/walkers/bqsr/RecalibrationArgumentCollection.java
@@ -153,6 +153,11 @@ public class RecalibrationArgumentCollection {
     @Argument(fullName = "quantizing_levels", shortName = "ql", required = false, doc = "number of distinct quality scores in the quantized output")
     public int QUANTIZING_LEVELS = 16;
 
+    /**
+     * The tag name for the binary tag covariate (if using it)
+     */
+    @Argument(fullName = "binary_tag_name", shortName = "bintag", required = false, doc = "the binary tag covariate name if using it")
+    public String BINARY_TAG_NAME = null;
 
     @Hidden
     @Argument(fullName = "default_platform", shortName = "dP", required = false, doc = "If a read has no platform then default to the provided String. Valid options are illumina, 454, and solid.")
@@ -205,6 +210,8 @@ public class RecalibrationArgumentCollection {
         argumentsTable.set("no_plots", RecalDataManager.ARGUMENT_VALUE_COLUMN_NAME, NO_PLOTS);
         argumentsTable.addRowID("recalibration_report", true);
         argumentsTable.set("recalibration_report", RecalDataManager.ARGUMENT_VALUE_COLUMN_NAME, recalibrationReport == null ? "null" : recalibrationReport.getAbsolutePath());
+        argumentsTable.addRowID("binary_tag_name", true);
+        argumentsTable.set("binary_tag_name", RecalDataManager.ARGUMENT_VALUE_COLUMN_NAME, BINARY_TAG_NAME == null ? "null" : BINARY_TAG_NAME);
         return argumentsTable;
     }
 
diff --git a/public/java/src/org/broadinstitute/sting/gatk/walkers/bqsr/RecalibrationReport.java b/public/java/src/org/broadinstitute/sting/gatk/walkers/bqsr/RecalibrationReport.java
index 9c6460320..e69cf4d69 100644
--- a/public/java/src/org/broadinstitute/sting/gatk/walkers/bqsr/RecalibrationReport.java
+++ b/public/java/src/org/broadinstitute/sting/gatk/walkers/bqsr/RecalibrationReport.java
@@ -290,6 +290,9 @@ public class RecalibrationReport {
 
             else if (argument.equals("recalibration_report"))
                 RAC.recalibrationReport = (value == null) ? null : new File((String) value);
+
+            else if (argument.equals("binary_tag_name"))
+                RAC.BINARY_TAG_NAME = (value == null) ? null : (String) value;
         }
 
         return RAC;