Update label for mGapReleaseAlleleFreqStep

Author: bbimber

mGAP/src/org/labkey/mgap/pipeline/mGapReleaseAlleleFreqStep.java

@@ -39,7 +39,7 @@ public static class Provider extends AbstractVariantProcessingStepProvider<mGapR
     {
         public Provider()
         {
-            super("mGapReleaseAlleleFreq", "Compare VCF to mGap Release", "DiscvrVariantAnnotator", "Annotate a VCF using the AF field from an mGAP release.", List.of(
+            super("mGapReleaseAlleleFreq", "Annotate using mGap Release AFs", "DiscvrVariantAnnotator", "Annotate a VCF using the AF field from an mGAP release.", List.of(
                     ToolParameterDescriptor.createExpDataParam(REF_VCF, "mGAP Release", "The mGAP release VCF to use for annotation", "sequenceanalysis-sequenceoutputfileselectorfield", new JSONObject()
                     {{
                         put("allowBlank", false);

mGAP/src/org/labkey/mgap/pipeline/mGapReleaseGenerator.java

@@ -1141,9 +1141,12 @@ private void inspectAndSummarizeVcf(JobContext ctx, File vcfInput, GeneToNameTra
                 return;
             }
 
+            String releaseVersion = ctx.getParams().optString("releaseVersion");
+
             long sitesInspected = 0L;
             long totalVariants = 0L;
             long totalPrivateVariants = 0L;
+            long newInThisRelease = 0L;
             Map<VariantContext.Type, Long> typeCounts = new HashMap<>();
 
             File interestingVariantTable = getVariantTableName(ctx, vcfInput);
@@ -1412,6 +1415,11 @@ else if (maxScore > 0.0)
                         }
                     }
 
+                    if (vc.getAttribute("mGAPV") != null && releaseVersion.equals(vc.getAttributeAsString("mGAPV", null)))
+                    {
+                        newInThisRelease++;
+                    }
+
                     for (List<String> line : queuedLines)
                     {
                         writer.writeNext(line.toArray(new String[0]));
@@ -1431,7 +1439,7 @@ else if (maxScore > 0.0)
                     totalSubjects = reader.getFileHeader().getSampleNamesInOrder().size();
                 }
 
-                generateSummaries(ctx, vcfInput, genome, totalVariants, totalPrivateVariants, totalSubjects, typeCounts);
+                generateSummaries(ctx, vcfInput, genome, totalVariants, totalPrivateVariants, newInThisRelease, totalSubjects, typeCounts);
             }
 
             try
@@ -1469,7 +1477,7 @@ public Collection<String> parseRawOmimPheno(VariantContext vc, Logger log)
             return Collections.emptySet();
         }
 
-        private static void generateSummaries(JobContext ctx, File vcf, ReferenceGenome genome, long totalVariants, long totalPrivateVariants, int totalSubjects, Map<VariantContext.Type, Long> typeCounts) throws PipelineJobException
+        private static void generateSummaries(JobContext ctx, File vcf, ReferenceGenome genome, long totalVariants, long totalPrivateVariants, long newInThisRelease, int totalSubjects, Map<VariantContext.Type, Long> typeCounts) throws PipelineJobException
         {
             //variants to table
             ctx.getLogger().info("Running VariantsToTable");
@@ -1504,7 +1512,7 @@ private static void generateSummaries(JobContext ctx, File vcf, ReferenceGenome
             ctx.getLogger().info("Generating summary stats from: " + variantsToTable.getName());
             File summaryTable = new File(vcf.getParentFile(), SequenceAnalysisService.get().getUnzippedBaseName(vcf.getName()) + ".summary.txt");
             File summaryTableByField = new File(vcf.getParentFile(), SequenceAnalysisService.get().getUnzippedBaseName(vcf.getName()) + ".summaryByField.txt");
-            new mGapSummarizer().generateSummary(ctx, variantsToTable, summaryTable, summaryTableByField, totalVariants, totalPrivateVariants, totalSubjects, typeCounts);
+            new mGapSummarizer().generateSummary(ctx, variantsToTable, summaryTable, summaryTableByField, totalVariants, totalPrivateVariants, newInThisRelease, totalSubjects, typeCounts);
         }
 
         private void maybeWriteVariantLine(Set<List<String>> queuedLines, VariantContext vc, @Nullable String allele, String source, String reason, String description, Collection<String> overlappingGenes, Collection<String> omimIds, Collection<String> omimPhenotypes, Logger log, String identifier)

mGAP/src/org/labkey/mgap/pipeline/mGapSummarizer.java

@@ -190,26 +190,24 @@ public static void filterCodingPotential(Set<String> codingPotential)
         }
     }
 
-    public void generateSummary(SequenceOutputHandler.JobContext ctx, File variantsToTable, File output, File outputPerValue, long totalVariants, long totalPrivateVariants, int totalSubjects, Map<VariantContext.Type, Long> typeCounts) throws PipelineJobException
+    public void generateSummary(SequenceOutputHandler.JobContext ctx, File variantsToTable, File output, File outputPerValue, long totalVariants, long totalPrivateVariants, long newInThisRelease, int totalSubjects, Map<VariantContext.Type, Long> typeCounts) throws PipelineJobException
     {
         ctx.getLogger().info("reading variant table");
-        int lineNo = 0;
         FieldTracker tracker = new FieldTracker(130);
         try (BufferedReader reader = Readers.getReader(variantsToTable))
         {
-            lineNo++;
-            if (lineNo % 500000 == 0)
-            {
-                ctx.getLogger().info("processed " + lineNo + " lines");
-            }
-
             String lineStr;
             List<String> header = new ArrayList<>();
             int lineCount = 0;
             while ((lineStr = reader.readLine()) != null)
             {
                 String[] line = lineStr.split("\t");
                 lineCount++;
+                if (lineCount % 500000 == 0)
+                {
+                    ctx.getLogger().info("processed " + lineCount + " lines");
+                }
+
                 if (lineCount == 1)
                 {
                     //skip header
@@ -249,6 +247,7 @@ public void generateSummary(SequenceOutputHandler.JobContext ctx, File variantsT
             valWriter.writeNext(new String[]{"Field", "Level", "Total"});
             valWriter.writeNext(new String[]{"Counts", "TotalVariants", String.valueOf(totalVariants)});
             valWriter.writeNext(new String[]{"Counts", "TotalPrivateVariants", String.valueOf(totalPrivateVariants)});
+            valWriter.writeNext(new String[]{"Counts", "TotalVariantsNewInRelease", String.valueOf(newInThisRelease)});
             valWriter.writeNext(new String[]{"Counts", "TotalSamples", String.valueOf(totalSubjects)});
 
             for (VariantContext.Type type : typeCounts.keySet())