From 28bf7ec8ad70ba8ba88bc649ed0efc571da5f1a3 Mon Sep 17 00:00:00 2001
From: kiran <kiran@348d0f76-0448-11de-a6fe-93d51630548a>
Date: Sun, 17 May 2009 04:09:23 +0000
Subject: [PATCH] Aesthetic cleanup.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@735 348d0f76-0448-11de-a6fe-93d51630548a
---
 .../secondarybase/AnnotateSecondaryBase.java  | 11 +++---
 .../sting/secondarybase/BasecallingStats.java | 35 +++++++++++++++----
 2 files changed, 35 insertions(+), 11 deletions(-)

diff --git a/java/src/org/broadinstitute/sting/secondarybase/AnnotateSecondaryBase.java b/java/src/org/broadinstitute/sting/secondarybase/AnnotateSecondaryBase.java
index c3117ec44..9886f9f72 100755
--- a/java/src/org/broadinstitute/sting/secondarybase/AnnotateSecondaryBase.java
+++ b/java/src/org/broadinstitute/sting/secondarybase/AnnotateSecondaryBase.java
@@ -36,7 +36,6 @@ public class AnnotateSecondaryBase extends CommandLineProgram {
     @Argument(fullName="cycle_end", shortName="CE", doc="On what cycle does the read end? (0-based inclusive)") public int CYCLE_END;
     @Argument(fullName="tlim", shortName="T", doc="Number of reads to use for parameter initialization", required=false) public int TRAINING_LIMIT = 250000;
     @Argument(fullName="clim", shortName="C", doc="Number of reads to basecall", required=false) public int CALLING_LIMIT = Integer.MAX_VALUE;
-    @Argument(fullName="context", shortName="X", doc="Attempt to correct for context?", required=false) public Boolean CONTEXT = false;
     @Argument(fullName="runbarcode", shortName="B", doc="Run barcode (embedded as part of the read name") public String RUN_BARCODE;
 
     public static void main(String[] argv) {
@@ -48,12 +47,12 @@ public class AnnotateSecondaryBase extends CommandLineProgram {
         BasecallingTrainingSet trainingSet = new BasecallingTrainingSet(BUSTARD_DIR, LANE, CYCLE_BEGIN, CYCLE_END, TRAINING_LIMIT);
 
         // Iterate through raw Firecrest data and store the first N reads up to TRAINING_LIMIT
-        System.out.println("Loading training set from the first " + TRAINING_LIMIT + " reads in the raw data...");
+        System.out.println("Loading training set from the first " + TRAINING_LIMIT + " unambiguous reads in the raw data...");
         trainingSet.loadFirstNUnambiguousReadsTrainingSet();
 
         // Iterate through the stored training data and add the info to the BasecallingReadModel
         System.out.println("Applying training set...");
-        BasecallingReadModel model = new BasecallingReadModel(CYCLE_END - CYCLE_BEGIN + 1, CONTEXT);
+        BasecallingReadModel model = new BasecallingReadModel(CYCLE_END - CYCLE_BEGIN + 1, true);
         model.train(trainingSet);
 
         // Call bases and write results
@@ -69,13 +68,15 @@ public class AnnotateSecondaryBase extends CommandLineProgram {
             FourProbRead fpr = model.call(rr);
 
             sfw.addAlignment(constructSAMRecord(rr, fpr, sfh, RUN_BARCODE));
-            bstats.update(rr, fpr, 10000);
+            bstats.update(rr, fpr);
+
+            bstats.notifyOnInterval(10000);
         }
 
         iparser.close();
         sfw.close();
 
-        System.out.printf("%% bases consistent: %d/%d (%4.4f)\n", bstats.getBasesConsistent(), bstats.getBasesTotal(), bstats.getPercentConsistent());
+        bstats.notifyNow();
         System.out.println("Done.");
 
         return 0;
diff --git a/java/src/org/broadinstitute/sting/secondarybase/BasecallingStats.java b/java/src/org/broadinstitute/sting/secondarybase/BasecallingStats.java
index a1edf0bb7..908f79c17 100755
--- a/java/src/org/broadinstitute/sting/secondarybase/BasecallingStats.java
+++ b/java/src/org/broadinstitute/sting/secondarybase/BasecallingStats.java
@@ -51,7 +51,7 @@ public class BasecallingStats {
      * @return the percent of bases called consistently
      */
     public double getPercentConsistent() {
-        return ((double) getBasesConsistent())/((double) getBasesTotal());
+        return 100.0*((double) getBasesConsistent())/((double) getBasesTotal());
     }
 
     /**
@@ -60,7 +60,7 @@ public class BasecallingStats {
      * @param rr   the raw Illumina read
      * @param fpr  the FourProb read
      */
-    public void update(RawRead rr, FourProbRead fpr, int updateInterval) {
+    public void update(RawRead rr, FourProbRead fpr) {
         for (int cycle = 0; cycle < fpr.size(); cycle++) {
             int rawBaseIndex = BaseUtils.simpleBaseToBaseIndex((char) rr.getSequence()[cycle]);
             int fpBaseIndex = fpr.get(cycle).indexAtRank(0);
@@ -74,10 +74,33 @@ public class BasecallingStats {
             }
         }
 
-        if (basesTotal % updateInterval == 0 && basesTotal > 0) {
-            System.out.printf("%% bases consistent: %d/%d (%4.4f)\r", basesConsistent, basesTotal, ((double) basesConsistent)/((double) basesTotal));
-        }
-
         readsTotal++;
     }
+
+    /**
+     * Returns basecalling stats info in a nicely formatted string
+     *
+     * @return nicely formatted string containing basecalling stats
+     */
+    public String toString() {
+        return String.format("%% bases consistent: %d/%d (%2.2f%%)", getBasesConsistent(), getBasesTotal(), getPercentConsistent());
+    }
+
+    /**
+     * Periodically print a line containing basecalling stats
+     *
+     * @param interval  the periodicity of the messages given in number of bases observed
+     */
+    public void notifyOnInterval(int interval) {
+        if (getBasesTotal() > 0 && getBasesTotal() % interval == 0) {
+            System.out.printf("%s\r", toString());
+        }
+    }
+
+    /**
+     * Immediately print a line containing basecalling stats
+     */
+    public void notifyNow() {
+        System.out.printf("%s\n", toString());
+    }
 }