Improve data validation

Author: bbimber

SivStudies/resources/queries/study/duplicatePVLs.query.xml

@@ -0,0 +1,12 @@
+<query xmlns="http://labkey.org/data/xml/query">
+    <metadata>
+        <tables xmlns="http://labkey.org/data/xml">
+            <table tableName="duplicatePVLs" tableDbType="NOT_IN_DB">
+                <tableTitle>Duplicate PVLs</tableTitle>
+                <columns>
+
+                </columns>
+            </table>
+        </tables>
+    </metadata>
+</query>

SivStudies/resources/queries/study/duplicatePVLs.sql

@@ -0,0 +1,12 @@
+SELECT
+t.Id,
+t.date,
+t.sampleType,
+t.assayType,
+t.target,
+t.result,
+t.dataSource
+
+FROM study.viralLoads t
+
+WHERE t.lsid IN (SELECT lsid FROM study.viralLoads v GROUP BY v.Id, v.date, v.sampleType, v.assayType HAVING count(*) > 1)

SivStudies/src/org/labkey/sivstudies/etl/PerformManualIdrStepsTask.java

@@ -95,6 +95,7 @@ private void pruneSivChallenges(PipelineJob pipelineJob) throws PipelineJobExcep
     private static final Pattern mir126_RE = Pattern.compile("miR[- ]{0,1}126",  Pattern.CASE_INSENSITIVE);
     private static final Pattern mir142_RE = Pattern.compile("miR[- ]{0,1}142",  Pattern.CASE_INSENSITIVE);
     private static final Pattern mir126_142_RE = Pattern.compile("miR[- ]{0,1}142[ ,-]{1,2}126",  Pattern.CASE_INSENSITIVE);
+    private static final Pattern pp71_RE = Pattern.compile("pp71",  Pattern.CASE_INSENSITIVE);
 
     private void updateVaccineInformation(PipelineJob pipelineJob) throws PipelineJobException
     {
@@ -133,6 +134,10 @@ else if (treatment.contains("RhCMV") && treatment.contains("d186-189"))
                 {
                     updatedRow.put("backbone", "68-1 d186-189");
                 }
+                else if (treatment.contains("RhCMV") && treatment.contains("pp71"))
+                {
+                    updatedRow.put("backbone", "68-1 delta-pp71");
+                }
             }
 
             if (treatment.toUpperCase().contains("MOCK"))