From 05d5e8be041de5c652535807fd3582ecf81791f5 Mon Sep 17 00:00:00 2001
From: jryurkanin <104916543+jryurkanin@users.noreply.github.com>
Date: Tue, 14 Feb 2023 09:11:57 -0800
Subject: [PATCH 1/7] Create parentDifferentSpecies.sql

identify animals with parents of a different species
---
 .../queries/study/parentDifferentSpecies.sql      | 15 +++++++++++++++
 1 file changed, 15 insertions(+)
 create mode 100644 ehr/resources/queries/study/parentDifferentSpecies.sql

diff --git a/ehr/resources/queries/study/parentDifferentSpecies.sql b/ehr/resources/queries/study/parentDifferentSpecies.sql
new file mode 100644
index 000000000..a7062e34c
--- /dev/null
+++ b/ehr/resources/queries/study/parentDifferentSpecies.sql
@@ -0,0 +1,15 @@
+SELECT * FROM (
+SELECT d1.Id,
+d1.Dataset.Demographics.gender,
+d1.Dataset.Demographics.species,
+d1.Parents.dam,
+d1.Parents.dam.Demographics.species as damSpecies,
+d1.Parents.sire,
+d1.Parents.sire.Demographics.species as sireSpecies,
+CASE WHEN (d1.Parents.dam.Demographics.species IS NOT NULL AND d1.Parents.dam.Demographics.species != d1.Dataset.Demographics.species) OR (d1.Parents.sire.Demographics.species IS NOT NULL AND d1.Parents.sire.Demographics.species != d1.Dataset.Demographics.species)
+THEN TRUE
+ELSE FALSE
+END as parentSpeciesMismatch
+FROM study.Animal d1
+) d2
+WHERE d2.parentSpeciesMismatch = TRUE

From f1a271d48c8e560680018cee5a9f809ac88d0aaf Mon Sep 17 00:00:00 2001
From: jryurkanin <104916543+jryurkanin@users.noreply.github.com>
Date: Tue, 14 Feb 2023 09:24:49 -0800
Subject: [PATCH 2/7] Create parentsWrongGender.sql

identify animals with a parent whose gender does not match their sire/dam designation
---
 .../queries/study/parentsWrongGender.sql          | 15 +++++++++++++++
 1 file changed, 15 insertions(+)
 create mode 100644 ehr/resources/queries/study/parentsWrongGender.sql

diff --git a/ehr/resources/queries/study/parentsWrongGender.sql b/ehr/resources/queries/study/parentsWrongGender.sql
new file mode 100644
index 000000000..7935cda24
--- /dev/null
+++ b/ehr/resources/queries/study/parentsWrongGender.sql
@@ -0,0 +1,15 @@
+SELECT * FROM (
+SELECT d1.Id,
+d1.Dataset.Demographics.gender,
+d1.Dataset.Demographics.species,
+d1.Parents.dam,
+d1.Parents.dam.Demographics.gender as damSex,
+d1.Parents.sire,
+d1.Parents.sire.Demographics.gender as sireSex,
+CASE WHEN (d1.Parents.dam.Demographics.gender.code IS NOT NULL AND d1.Parents.dam.Demographics.gender.code != 'f') OR (d1.Parents.sire.Demographics.gender.code IS NOT NULL AND d1.Parents.sire.Demographics.gender.code != 'm')
+THEN TRUE
+ELSE FALSE
+END as parentSexMismatch
+FROM study.Animal d1
+) d2
+WHERE d2.parentSexMismatch = TRUE

From 3e35ccb15fd2ffd71255d7af229437df92af2e67 Mon Sep 17 00:00:00 2001
From: jryurkanin <104916543+jryurkanin@users.noreply.github.com>
Date: Tue, 14 Feb 2023 09:40:19 -0800
Subject: [PATCH 3/7] Create parentsYoungerThanOffspring.sql

identifies animals whose parents that are younger than their offspring
---
 .../queries/study/parentsYoungerThanOffspring.sql | 15 +++++++++++++++
 1 file changed, 15 insertions(+)
 create mode 100644 ehr/resources/queries/study/parentsYoungerThanOffspring.sql

diff --git a/ehr/resources/queries/study/parentsYoungerThanOffspring.sql b/ehr/resources/queries/study/parentsYoungerThanOffspring.sql
new file mode 100644
index 000000000..76bc3589c
--- /dev/null
+++ b/ehr/resources/queries/study/parentsYoungerThanOffspring.sql
@@ -0,0 +1,15 @@
+SELECT * FROM (
+SELECT 
+dem.Id,
+dem.gender,
+dem.species,
+dem.birth,
+dem.dam,
+damDem.birth as damBirth,
+dem.sire,
+sireDem.birth as sireBirth
+FROM demographics dem
+LEFT JOIN demographics damDem ON dem.dam = damDem.Id
+LEFT JOIN demographics sireDem ON dem.sire = sireDem.Id
+   ) t
+WHERE (t.birth <= t.damBirth OR t.birth <= t.sireBirth)

From fc456d7f8a4d8c78c42486c35a312ec8152ddfe4 Mon Sep 17 00:00:00 2001
From: jryurkanin <104916543+jryurkanin@users.noreply.github.com>
Date: Sun, 26 Feb 2023 11:27:06 -0800
Subject: [PATCH 4/7] Rename parentDifferentSpecies.sql to
 parentsDifferentSpecies.sql

---
 .../{parentDifferentSpecies.sql => parentsDifferentSpecies.sql}   | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 rename ehr/resources/queries/study/{parentDifferentSpecies.sql => parentsDifferentSpecies.sql} (100%)

diff --git a/ehr/resources/queries/study/parentDifferentSpecies.sql b/ehr/resources/queries/study/parentsDifferentSpecies.sql
similarity index 100%
rename from ehr/resources/queries/study/parentDifferentSpecies.sql
rename to ehr/resources/queries/study/parentsDifferentSpecies.sql

From c59a4487f4b4075c9c364558e23d992329a73155 Mon Sep 17 00:00:00 2001
From: jryurkanin <104916543+jryurkanin@users.noreply.github.com>
Date: Sun, 26 Feb 2023 11:29:59 -0800
Subject: [PATCH 5/7] create parentsBothDamAndSire.sql

identify animals which are listed as both a dam and a sire
---
 .../queries/study/parentsBothDamAndSire.sql   | 30 +++++++++++++++++++
 1 file changed, 30 insertions(+)
 create mode 100644 ehr/resources/queries/study/parentsBothDamAndSire.sql

diff --git a/ehr/resources/queries/study/parentsBothDamAndSire.sql b/ehr/resources/queries/study/parentsBothDamAndSire.sql
new file mode 100644
index 000000000..b7b34dc1e
--- /dev/null
+++ b/ehr/resources/queries/study/parentsBothDamAndSire.sql
@@ -0,0 +1,30 @@
+SELECT t.Id as Parent, GROUP_CONCAT(t.DamChildren, ',') as DamChildren, GROUP_CONCAT(t.SireChildren, ',') as SireChildren FROM
+(
+
+SELECT d.Id, d.DamChildren, s.SireChildren FROM (
+
+SELECT Id as DamChildren,
+dam as Id
+FROM demographics
+WHERE dam IN (
+SELECT Distinct dam as Id 
+FROM demographics
+WHERE dam IN (SELECT Distinct sire FROM demographics))
+) d   
+
+JOIN
+
+(SELECT * FROM (
+
+SELECT Id as SireChildren,
+sire as Id
+FROM demographics
+WHERE sire IN (
+SELECT Distinct dam as Id 
+FROM demographics
+WHERE dam IN (SELECT Distinct sire FROM demographics))
+) si ) s   
+   
+ON s.Id = d.Id   
+) t
+GROUP BY t.Id
\ No newline at end of file

From 005f83b58d2a3a42df63c5717014ca837eaa8a98 Mon Sep 17 00:00:00 2001
From: jryurkanin <104916543+jryurkanin@users.noreply.github.com>
Date: Sun, 26 Feb 2023 11:36:29 -0800
Subject: [PATCH 6/7] create dataValidationReport page

---
 ehr/resources/views/dataValidationReport.html | 71 +++++++++++++++++++
 .../views/dataValidationReport.view.xml       |  5 ++
 2 files changed, 76 insertions(+)
 create mode 100644 ehr/resources/views/dataValidationReport.html
 create mode 100644 ehr/resources/views/dataValidationReport.view.xml

diff --git a/ehr/resources/views/dataValidationReport.html b/ehr/resources/views/dataValidationReport.html
new file mode 100644
index 000000000..32b4d0b39
--- /dev/null
+++ b/ehr/resources/views/dataValidationReport.html
@@ -0,0 +1,71 @@
+<p>
+    Each of the below data grids identifies issues present in the animal colony data that would prevent the pedigree plots from working correctly.<br>
+    To correct any data consistency issues, use this report as a reference for identifying the data issues, and then correct them on the demographics dataset, which can be access with the below button.<br>
+</p>
+<button onclick = "linkBrowseData()">
+    Correct Data
+</button>
+<p><br></p>
+<table>
+    <tr>
+        <td>
+            <div name="webpart" id='parentsDifferentSpeciesReport' />
+        </td>
+        <td>
+            <div name="webpart" id='parentsBothDamAndSireReport' />
+        </td>
+    </tr>
+    <tr>
+        <td>
+            <div name="webpart" id='parentsWrongGenderReport' />
+        </td>
+        <td>
+            <div name="webpart" id='parentsYoungerThanOffspringReport' />
+        </td>
+    </tr>
+</table>
+
+<script>
+
+function linkBrowseData() {
+    window.location = LABKEY.ActionURL.buildURL(
+        "ehr",
+        "updateQuery",
+        LABKEY.ActionURL.getContainer(),
+        {
+            schemaName: "study",
+            queryName: "demographics",
+            returnUrl: window.location
+        }
+        );
+}
+
+new LABKEY.QueryWebPart({
+            renderTo: 'parentsDifferentSpeciesReport',
+            title: 'Parents that are wrong species',
+            schemaName: 'study',
+            queryName: 'parentsDifferentSpecies'
+});
+
+new LABKEY.QueryWebPart({
+            renderTo: 'parentsBothDamAndSireReport',
+            title: 'Parents that are listed as both dam and sire',
+            schemaName: 'study',
+            queryName: 'parentsBothDamAndSire'
+});
+
+new LABKEY.QueryWebPart({
+            renderTo: 'parentsWrongGenderReport',
+            title: 'Parents that are listed as both dam and sire',
+            schemaName: 'study',
+            queryName: 'parentsWrongGender'
+});
+
+new LABKEY.QueryWebPart({
+            renderTo: 'parentsYoungerThanOffspringReport',
+            title: 'Parents that are listed as both dam and sire',
+            schemaName: 'study',
+            queryName: 'parentsYoungerThanOffspring'
+});
+
+</script>
\ No newline at end of file
diff --git a/ehr/resources/views/dataValidationReport.view.xml b/ehr/resources/views/dataValidationReport.view.xml
new file mode 100644
index 000000000..0f0b3dc97
--- /dev/null
+++ b/ehr/resources/views/dataValidationReport.view.xml
@@ -0,0 +1,5 @@
+<view xmlns="http://labkey.org/data/xml/view" frame="none" title="Data Validation Report">
+    <dependencies>
+        <dependency path="ehr.context" />
+    </dependencies>
+</view>
\ No newline at end of file

From e4f92562c6979540b8528f885480855a6283fd7d Mon Sep 17 00:00:00 2001
From: jryurkanin <104916543+jryurkanin@users.noreply.github.com>
Date: Sun, 26 Feb 2023 11:50:24 -0800
Subject: [PATCH 7/7] add dataValidationReport to nav menu

---
 ehr/resources/views/begin.html | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/ehr/resources/views/begin.html b/ehr/resources/views/begin.html
index 4739d1577..93c9a679c 100644
--- a/ehr/resources/views/begin.html
+++ b/ehr/resources/views/begin.html
@@ -86,6 +86,7 @@
             {header: 'Misc',
             items: [
                 {name: 'About the EHR Project', url: '<%=contextPath%>/project' + ctx['EHRStudyContainer'] + '/About The EHR/begin.view?'},
+                {name: 'Data Validation', url: '<%=contextPath%>/ehr' + ctx['EHRStudyContainer'] + '/dataValidationReport.view?'},
                 {name: 'Compare Lists of Animals', url: '<%=contextPath%>/ehr' + ctx['EHRStudyContainer'] + '/utilities.view?'},
                 {name: 'Run SQL Directly', url: '<%=contextPath%>/ehr' + ctx['EHRStudyContainer'] + '/executeSql.view?'},
                 {name: 'Search Center SNOMED Codes', url: '<%=contextPath%>/query' + ctx['EHRStudyContainer'] + '/executeQuery.view?schemaName=ehr_lookups&query.queryName=snomed'},
@@ -120,4 +121,4 @@
 
 }
 
-</script>
\ No newline at end of file
+</script>