Fix for #424: move STARSoloChimericReadFilteringIterator down

so reads without UMIs don't reach it.

src/java/org/broadinstitute/dropseqrna/barnyard/digitalallelecounts/SNPUMIBasePileupIterator.java

@@ -98,21 +98,21 @@ public SNPUMIBasePileupIterator (final SamHeaderAndIterator headerAndIter, final
 
         // add a progress logger.
         final ProgressLoggingIterator loggingIterator = new ProgressLoggingIterator (headerAndIter.iterator, logger);
-		final STARSoloChimericReadFilteringIterator chimericReadFilteringIterator = new STARSoloChimericReadFilteringIterator(loggingIterator, molecularBarcodeTag);
 
         // filter out to contigs in the interval list.
         Set<String> contigsToFilter = snpIntervals.getIntervals().stream().map(x -> x.getContig()).collect(Collectors.toSet());
-        ChromosomeFilteringIterator cfi = new ChromosomeFilteringIterator(chimericReadFilteringIterator, contigsToFilter, false);
+        ChromosomeFilteringIterator cfi = new ChromosomeFilteringIterator(loggingIterator, contigsToFilter, false);
 
         // Filter reads on map quality
      	MapQualityFilteredIterator filteringIterator = new MapQualityFilteredIterator(cfi, readMQ, true);
 
         // Filter records before sorting, to reduce I/O
         final MissingTagFilteringIterator filteringIterator2 =
                 new MissingTagFilteringIterator(filteringIterator, geneTag, cellBarcodeTag, molecularBarcodeTag);
-
+
+		final STARSoloChimericReadFilteringIterator chimericReadFilteringIterator = new STARSoloChimericReadFilteringIterator(filteringIterator2, molecularBarcodeTag);
      	// Filter/assign reads based on functional annotations
-     	GeneFunctionIteratorWrapper gfteratorWrapper = new GeneFunctionIteratorWrapper(filteringIterator2, geneTag, geneStrandTag, geneFunctionTag, assignReadsToAllGenes, strandStrategy, acceptedLociFunctions, functionStrategy);
+     	GeneFunctionIteratorWrapper gfteratorWrapper = new GeneFunctionIteratorWrapper(chimericReadFilteringIterator, geneTag, geneStrandTag, geneFunctionTag, assignReadsToAllGenes, strandStrategy, acceptedLociFunctions, functionStrategy);
 
      	/**
      	 * In this section, data is sorted into the standard cell / gene / umi order and grouped so that all reads for a single UMI can be evaluated at once.

src/tests/java/org/broadinstitute/dropseqrna/barnyard/digitalallelecounts/SNPUMIBasePileupIteratorTest.java

@@ -89,7 +89,26 @@ public void testPileups() {
 		}
 		Assert.assertEquals(expectedPileups, count);
 	}
-
+
+	@Test
+	public void testMissingUMITag() {
+		List<String> cellBarcodes = ParseBarcodeFile.readCellBarcodeFile(cellBCFile);
+		IntervalList snpIntervals = IntervalList.fromFile(snpIntervalsFile);
+
+		// all SNPs equal quality.  No SNPs overlap on UMIs so all pileups are seen.
+		Map<Interval, Double> meanGenotypeQuality = new HashMap<Interval, Double>();
+		for (Interval i: snpIntervals) {
+			meanGenotypeQuality.put(i, 30d);
+		}
+
+		SNPUMIBasePileupIterator sbpi = new SNPUMIBasePileupIterator(
+				new SamHeaderAndIterator(smallBAMFile), snpIntervals, GENE_NAME_TAG, GENE_STRAND_TAG, GENE_FUNCTION_TAG,
+				LOCUS_FUNCTION_LIST, STRAND_STRATEGY, GeneFunctionCommandLineBase.DEFAULT_FUNCTIONAL_STRATEGY, cellBarcodeTag,
+				"XQ", snpTag, functionTag, readMQ, assignReadsToAllGenes, cellBarcodes, meanGenotypeQuality,
+				SortOrder.SNP_GENE);
+		Assert.assertFalse(sbpi.hasNext());
+
+	}
 
 	// for reads that have more than 1 SNP in the data, filter the SNPs tagging the read to a single SNP.
 	// This means the read only contributes to a single pileup.