Fixing output printing issues in the code, as well as adding more safety checks

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@105 348d0f76-0448-11de-a6fe-93d51630548a

core/java/src/org/broadinstitute/sting/gatk/TraversalEngine.java

@@ -237,12 +237,6 @@ public class TraversalEngine {
      * @param loc Current location 
      */
     public void printProgress( boolean mustPrint, final String type, GenomeLoc loc ) {
-        // If an index is enabled, file read progress is meaningless because a linear
-        // traversal is not being performed.  For now, don't bother printing progress.
-        // TODO: Create a sam indexed read tracker that tracks based on percentage through the query.
-        if( samReadingTracker == null )
-            return;
-
         final long nRecords = this.nRecords;
         final long curTime = System.currentTimeMillis();
         final double elapsed = (curTime - startTime) / 1000.0;
@@ -257,7 +251,11 @@ public class TraversalEngine {
                 System.out.printf("[PROGRESS] Traversed %,d %s in %.2f secs (%.2f secs per 1M %s)%n", nRecords, type, elapsed, secsPer1MReads, type);
 
             // Currently samReadingTracker will print misleading info if we're not processing the whole file
-            if ( this.locs == null )
+
+            // If an index is enabled, file read progress is meaningless because a linear
+            // traversal is not being performed.  For now, don't bother printing progress.
+            // TODO: Create a sam indexed read tracker that tracks based on percentage through the query.
+            if ( samReadingTracker != null && this.locs == null )
                 System.out.printf("[PROGRESS]   -> %s%n", samReadingTracker.progressMeter());
         }
     }

core/java/src/org/broadinstitute/sting/gatk/iterators/ReferenceIterator.java

@@ -112,7 +112,13 @@ public class ReferenceIterator implements Iterator<ReferenceIterator> {
             next();
 
         //System.out.printf("  -> Seeking to %s %d from %s %d%n", contigName, seekOffset, currentContig.getName(), offset);
-        if ( contigName.equals(currentContig.getName()) ) {
+        int cmpContigs = GenomeLoc.compareContigs(contigName, currentContig.getName());
+        if ( cmpContigs == -1 ) {
+            // The contig we are looking for is before the currentContig -- it's an error
+            throw new IllegalArgumentException(String.format("Invalid seek to %s from %s, which is usually due to out of order reads%n",
+                    new GenomeLoc(currentContig.getName(), seekOffset), new GenomeLoc(currentContig.getName(), offset)));
+        }
+        else if ( cmpContigs == 0 ) {
             // we're somewhere on this contig
             if ( seekOffset < offset || seekOffset >= currentContig.length() ) {
                 // bad boy -- can't go backward safely or just beyond the contig length
@@ -134,7 +140,7 @@ public class ReferenceIterator implements Iterator<ReferenceIterator> {
                     // never found anything
                     return null;
                 }
-                else if ( nextContig.getName().equals(contigName) ) {
+                else if ( GenomeLoc.compareContigs( nextContig.getName(), contigName ) == 0 ) {
                     swapNextContig();
                     return seekForward(contigName, seekOffset);
                 }

core/java/src/org/broadinstitute/sting/gatk/iterators/ThreadedIterator.java

@@ -16,6 +16,7 @@ public class ThreadedIterator<T> implements Iterator<T>, Runnable {
     private Iterator<T> it;
     private final BlockingQueue<T> queue;
     private int nOps = 0;
+    private final int printStateFreq = -1;
 
     public void run() {
         try {
@@ -30,7 +31,7 @@ public class ThreadedIterator<T> implements Iterator<T>, Runnable {
     }
 
     public synchronized void printState(final String op) {
-        if ( nOps++ % 100000 == 0 )
+        if ( printStateFreq != -1 && nOps++ % printStateFreq == 0 )
             System.out.printf("  [%s] Queue has %d elements %d ops%n", op, queue.size(), nOps);
     }
 

playground/python/StressTestGATK.py

@@ -3,36 +3,42 @@ import os.path
 import sys
 import getopt
 
+defaultCommands = ['CountLoci', 'Pileup']
+
 def usage():
     print "Optional arguments:"
     print "  -f QUEUE   Farm jobs to QUEUE on LSF"
-
+    print "  -c cmd1,cmd2   Walkers to execute, otherwise", ' '.join(defaultCommands)
+    
 if __name__ == "__main__":
     opts = None
     try:
-        opts, args = getopt.getopt(sys.argv[1:], "f:", ["farm"])
+        opts, args = getopt.getopt(sys.argv[1:], "f:c:", ["farm", "commands"])
     except getopt.GetoptError:
         print sys.argv
         usage()
         sys.exit(2)
 
     farm_sub = False
+    commandsList = defaultCommands
 
     for opt, arg in opts:
         if opt in ("-f", "--farm"):
             farm_sub = arg
+        if opt in ("-c", "--commands"):
+            commandsList = arg.split(',')
 
-    for line in open(sys.argv[1]):
+    for line in open(args[0]):
         lane = line.strip()
         head, lane_filename = os.path.split(lane)
         filebase = os.path.splitext(lane_filename)[0]
     
         # convert the fasta
-        for analysis in ['CountLoci', 'Pileup']:
+        for analysis in commandsList:
             output = filebase + '.' + analysis + '.output'
-            if not os.path.exists(output):
-                cmd = "java -jar ~/dev/GenomeAnalysisTK/trunk/playground/java/dist/GenomeAnalysisTK.jar T=" + analysis + " I= " + lane + " R= /seq/references/Homo_sapiens_assembly18/v0/Homo_sapiens_assembly18.fasta"
-                print cmd
-                farm_commands.cmd(cmd, farm_sub, output, just_print_commands=True)
+            #if not os.path.exists(output):
+            cmd = "java -jar ~/dev/GenomeAnalysisTK/trunk/playground/java/dist/GenomeAnalysisTK.jar T=" + analysis + " I= " + lane + " R= /seq/references/Homo_sapiens_assembly18/v0/Homo_sapiens_assembly18.fasta "
+            print cmd
+            farm_commands.cmd(cmd, farm_sub, output)