Expose option to change strandedness threshold

Author: bbimber

SequenceAnalysis/src/org/labkey/sequenceanalysis/analysis/AbstractCombineGeneCountsHandler.java

@@ -81,11 +81,16 @@ private static List<ToolParameterDescriptor> getParams(boolean allowInferStrande
 
         if (allowInferStranded)
         {
-            ret.add(ToolParameterDescriptor.create(STRANDED, "Strand 1", "Choose whether to treat these data as stranded, unstranded, or to have the script infer the strandedness", "ldk-simplecombo", new JSONObject()
+            ret.add(ToolParameterDescriptor.create(STRANDED, "Strandedness", "Choose whether to treat these data as stranded, unstranded, or to have the script infer the strandedness", "ldk-simplecombo", new JSONObject()
             {{
                 put("storeValues", STRAND1 + ";" + STRAND2 + ";" + UNSTRANDED + ";" + INFER);
                 put("value", INFER);
             }}, INFER));
+            ret.add(ToolParameterDescriptor.create("strandedThreshold", "Strandedness Threshold", "If infer is selected, Choose whether to treat these data as stranded, unstranded, or to have the script infer the strandedness", "ldk-simplecombo", new JSONObject(){{
+                put("minValue", 0);
+                put("maxValue", 1);
+                put("decimalPrecision", 2);
+            }}, 0.9));
         }
 
         return ret;
@@ -305,29 +310,20 @@ public void prepareFiles(JobContext ctx, List<SequenceOutputFile> inputFiles, St
                 header.add(hp.getHeader(ctx, outputFileMap.get(rowId)));
             }
 
-            writer.writeNext(header.toArray(new String[header.size()]));
+            writer.writeNext(header.toArray(new String[0]));
 
             Set<String> genesWithoutData = new TreeSet<>();
             for (String geneId : results.distinctGenes)
             {
                 List<String> row = new ArrayList<>(inputFiles.size() + 3);
                 if (translator.getGeneMap().containsKey(geneId))
                 {
-                    if (translator.getGeneMap().containsKey(geneId))
-                    {
-                        row.add(geneId);
-                        row.add(translator.getGeneMap().get(geneId).get("gene_name"));
-                        row.add(translator.getGeneMap().get(geneId).get("gene_description"));
-                    }
-                    else
-                    {
-                        row.add(geneId);
-                        row.add("");
-                        row.add("");
-                    }
+                    row.add(geneId);
+                    row.add(translator.getGeneMap().get(geneId).get("gene_name"));
+                    row.add(translator.getGeneMap().get(geneId).get("gene_description"));
 
                     List<String> toAdd = new ArrayList<>();
-                    Integer totalWithData = 0;
+                    int totalWithData = 0;
                     for (Integer rowId : outputFileMap.keySet())
                     {
                         Double count = results.counts.get(rowId).get(geneId);
@@ -341,9 +337,9 @@ public void prepareFiles(JobContext ctx, List<SequenceOutputFile> inputFiles, St
 
                     if (totalWithData > 0 || !doSkipGenesWithoutData)
                     {
-                        row.add(totalWithData.toString());
+                        row.add(String.valueOf(totalWithData));
                         row.addAll(toAdd);
-                        writer.writeNext(row.toArray(new String[row.size()]));
+                        writer.writeNext(row.toArray(new String[0]));
                     }
 
                     if (totalWithData == 0 && !OTHER_IDS.contains(geneId))

SequenceAnalysis/src/org/labkey/sequenceanalysis/analysis/CombineStarGeneCountsHandler.java

@@ -31,7 +31,6 @@ public CombineStarGeneCountsHandler()
     @Override
     protected void processOutputFiles(CountResults results, List<SequenceOutputFile> inputFiles, JSONObject params, GeneToNameTranslator translator, PipelineJob job, RecordedAction action) throws PipelineJobException
     {
-
         String strandedSelection = params.optString(STRANDED, INFER);
 
         //next iterate all read count TSVs to guess strandedness
@@ -69,12 +68,12 @@ protected void processOutputFiles(CountResults results, List<SequenceOutputFile>
                         continue;
                     }
 
-                    Long unstranded = Long.parseLong(cells[1]);
-                    Long strand1 = Long.parseLong(cells[2]);
+                    long unstranded = Long.parseLong(cells[1]);
+                    long strand1 = Long.parseLong(cells[2]);
                     totalStrand1 += strand1;
-                    Long strand2 = Long.parseLong(cells[3]);
+                    long strand2 = Long.parseLong(cells[3]);
                     totalStrand2 += strand2;
-                    Long strandMax = Math.max(strand1, strand2);
+                    long strandMax = Math.max(strand1, strand2);
 
                     results.distinctGenes.add(geneId);
 
@@ -103,9 +102,9 @@ protected void processOutputFiles(CountResults results, List<SequenceOutputFile>
                         strand2Map = new HashMap<>(Math.max(results.distinctGenes.size() + 500, 5000));
                     }
 
-                    unstrandedMap.put(geneId, unstranded.doubleValue());
-                    strand1Map.put(geneId, strand1.doubleValue());
-                    strand2Map.put(geneId, strand2.doubleValue());
+                    unstrandedMap.put(geneId, (double)unstranded);
+                    strand1Map.put(geneId, (double)strand1);
+                    strand2Map.put(geneId, (double)strand2);
 
                     unstrandedCounts.put(so.getRowid(), unstrandedMap);
                     strand1Counts.put(so.getRowid(), strand1Map);
@@ -125,8 +124,9 @@ protected void processOutputFiles(CountResults results, List<SequenceOutputFile>
 
         //finally build output
         double avgStrandRatio = sumStrandRatio / countStrandRatio;
+        double threshold = params.optDouble("strandedThreshold", 0.9);
         job.getLogger().info("the average stranded/unstranded ratio for all samples was: " + avgStrandRatio);
-        double threshold = 0.9;
+
         String inferredStrandedness;
         job.getLogger().info("Attempting to infer strandedness");
         if (avgStrandRatio > threshold)