Updated the sequence reader

Author: wangqion

nbproject/project.properties

@@ -91,4 +91,4 @@ run.test.classpath=\
 source.encoding=UTF-8
 src.dir=src
 test.src.dir=test
-project.license=gpl20
+project.license=gpl30

src/edu/msu/cme/rdp/classifier/Classifier.java

@@ -63,7 +63,7 @@ public ClassificationResult classify(Sequence seq) throws IOException {
         return classify(new ClassifierSequence(seq));
     }
 
-    public ClassificationResult classify(ClassifierSequence seq) throws IOException {
+    public ClassificationResult classify(ClassifierSequence seq) {
         return classify(seq, MIN_BOOTSTRSP_WORDS );
     }
 
@@ -74,15 +74,16 @@ public ClassificationResult classify(ClassifierSequence seq) throws IOException
      * the number of bootstrap trials was used as an estimate of confidence in the assignment to that genus.
      * @throws ShortSequenceException if the sequence length is less than the minimum sequence length.
      */
-    public ClassificationResult classify(ClassifierSequence seq, int min_bootstrap_words) throws IOException {
+    public ClassificationResult classify(ClassifierSequence seq, int min_bootstrap_words) {
         GenusWordConditionalProb gProb = null;
         int nodeListSize = trainingInfo.getGenusNodeListSize();
         boolean reversed = false;
 
-        int [] wordIndexArr = seq.createWordIndexArr();
+        try {
+        int [] wordIndexArr = seq.getWordIndexArr();
         if (trainingInfo.isSeqReversed(wordIndexArr, seq.getGoodWordCount())) {
             seq = seq.getReversedSeq();
-            wordIndexArr = seq.createWordIndexArr();
+            wordIndexArr = seq.getWordIndexArr();
             reversed = true;
         }
 
@@ -99,8 +100,7 @@ public ClassificationResult classify(ClassifierSequence seq, int min_bootstrap_w
         }
 
         if (goodWordCount > MAX_NUM_OF_WORDS) {
-            querySeq_wordProbArr = new float[goodWordCount][nodeListSize];
-            System.err.println("increase the array size to " + goodWordCount);
+            querySeq_wordProbArr = new float[goodWordCount][nodeListSize];            
         }
 
         int NUM_OF_SELECTIONS = Math.max( goodWordCount / GoodWordIterator.getWordsize(), min_bootstrap_words);
@@ -199,6 +199,10 @@ public ClassificationResult classify(ClassifierSequence seq, int min_bootstrap_w
         ClassificationResult finalResult = new ClassificationResult(seq, reversed, finalAssigns, trainingInfo.getHierarchyInfo());
 
         return finalResult;
+        } catch (IOException ex){
+            throw new ShortSequenceException(seq.getSeqName(), "ShortSequenceException: The length of sequence with recordID="
+                    + seq.getSeqName() + " is less than " + MIN_SEQ_LEN);
+        }
     }
 
     /**

src/edu/msu/cme/rdp/classifier/TrainingInfo.java

@@ -258,7 +258,7 @@ private void createGenusNodeList(HierarchyTree root) {
      * less that zero, the query sequence is in reverse orientation.
      */
     public boolean isSeqReversed(ClassifierSequence seq) throws IOException {
-        int[] wordIndexArr = seq.createWordIndexArr();
+        int[] wordIndexArr = seq.getWordIndexArr();
         boolean reverse = false;
         float priorDiff = 0;
         for (int offset = 0; offset < wordIndexArr.length; offset++) {

src/edu/msu/cme/rdp/classifier/utils/ClassifierSequence.java

@@ -22,16 +22,23 @@
 public class ClassifierSequence extends Sequence {
     private boolean reverse = false;
     private Integer goodWordCount = null; // the number of words with only valid bases
-
+    private int [] wordIndexArr = null; 
     /**
      * Creates new ParsedSequence.
      */
-    public ClassifierSequence(Sequence seq) {
+    public ClassifierSequence(Sequence seq) throws IOException{
         this(seq.getSeqName(), seq.getDesc(), seq.getSeqString());
     }
 
-    public ClassifierSequence(String seqName, String desc, String seqString) {
+    public ClassifierSequence(String seqName, String desc, String seqString) throws IOException {
         super(seqName, desc, SeqUtils.getUnalignedSeqString(seqString));
+        /**
+        * Fetches every overlapping word from the sequence string, changes each
+        * word to integer format and saves in an array.
+        */
+        GoodWordIterator iterator = new GoodWordIterator(this.getSeqString());
+        this.wordIndexArr = iterator.getWordArr();        
+        this.goodWordCount = wordIndexArr.length;
     }
 
     /**
@@ -41,6 +48,9 @@ protected void setSeqString(String s) {
         seqString = s;
     }
 
+    public int[] getWordIndexArr(){
+        return this.wordIndexArr;
+    }
     /**
      * Returns true if the sequence string is a minus strand.
      */
@@ -52,31 +62,16 @@ public boolean isReverse() {
      * Returns a Sequence object whose sequence string is the reverse complement
      * of the current rRNA sequence string.
      */
-    public ClassifierSequence getReversedSeq() {
+    public ClassifierSequence getReversedSeq() throws IOException {
         ClassifierSequence retval = new ClassifierSequence(seqName, desc, IUBUtilities.reverseComplement(seqString));
         retval.reverse = true;
         return retval;
     }
 
-    /**
-     * Fetches every overlapping word from the sequence string, changes each
-     * word to integer format and saves in an array.
-     */
-    public int[] createWordIndexArr() throws IOException {
-        GoodWordIterator iterator = new GoodWordIterator(this.getSeqString());
-        int [] wordIndexArr = iterator.getWordArr();
-        
-        this.goodWordCount = wordIndexArr.length;
-        return wordIndexArr;
-    }
-
     /**
      * Returns the number of words with valid bases.
      */
-    public int getGoodWordCount() throws IOException {
-        if (goodWordCount == null) {
-            this.createWordIndexArr();
-        }
+    public int getGoodWordCount() {
         return goodWordCount;
     }
 }

test/edu/msu/cme/rdp/classifier/rrnaclassifier/ParsedSequenceTest.java

@@ -38,7 +38,7 @@ public static Test suite() {
      * Test of getReversedSeq method, of class
      * edu.msu.cme.rdp.classifier.readseqwrapper.ParsedSequence.
      */
-    public void testGetReversedSeq() {
+    public void testGetReversedSeq() throws IOException {
         System.out.println("testGetReversedSeq");
 
         //                 "AAAAAAAAAG-CCCCCCCCUGAGGGUUACnAA";
@@ -107,7 +107,7 @@ public void testCreateWordIndexArr() throws IOException {
         String seqString = "AAAAAAAAAG-CCCCCCCCUGAGGGUUACnAA";
         ClassifierSequence aSeq = new ClassifierSequence("test", "", seqString);
 
-        int[] wordIndexArr = aSeq.createWordIndexArr();
+        int[] wordIndexArr = aSeq.getWordIndexArr();
 
         assertEquals(0, wordIndexArr[0]);   //AAAAAAAA