Update schema for donor metadata (#655)

LabKey · Jul 1, 2024 · 973358d · 973358d
1 parent 985e48c
commit 973358d
Show file tree

Hide file tree

Showing 28 changed files with 519 additions and 233 deletions.
diff --git a/resources/queries/cds/MAbMixMAbMeta.sql b/resources/queries/cds/MAbMixMAbMeta.sql
@@ -29,11 +29,13 @@ SELECT
   mab_hxb2_location,
   mab_ab_binding_type,
   mab_isotype,
-  mab_donorid,
-  mab_donor_species,
-  mab_donor_clade
+  mabmeta.donor_id AS mab_donorid,
+-- donor meta
+  donormeta.donor_species AS mab_donor_species,
+  donormeta.donor_clade AS mab_donor_clade
 
 FROM MAbMix mix
 
 LEFT JOIN MAbMixMetadata mixmeta ON mix.mab_mix_id = mixmeta.mab_mix_id
-LEFT JOIN MAbMetadata mabmeta ON mix.mab_id = mabmeta.mab_id
+LEFT JOIN MAbMetadata mabmeta ON mix.mab_id = mabmeta.mab_id
+LEFT JOIN donor_metadata donormeta ON mabmeta.donor_id = donormeta.donor_id
diff --git a/resources/queries/cds/MAbMixMAbMetaForLearnExport.sql b/resources/queries/cds/MAbMixMAbMetaForLearnExport.sql
@@ -32,11 +32,13 @@ SELECT
   mab_hxb2_location,
   mab_ab_binding_type,
   mab_isotype,
-  mab_donorid,
-  mab_donor_species,
-  mab_donor_clade
+  mabmeta.donor_id AS mab_donorid,
+-- donor meta
+  donormeta.donor_species AS mab_donor_species,
+  donormeta.donor_clade AS mab_donor_clade
 
 FROM MAbMix mix
 
 LEFT JOIN MAbMixMetadata mixmeta ON mix.mab_mix_id = mixmeta.mab_mix_id
-LEFT JOIN MAbMetadata mabmeta ON mix.mab_id = mabmeta.mab_id
+LEFT JOIN MAbMetadata mabmeta ON mix.mab_id = mabmeta.mab_id
+LEFT JOIN donor_metadata donormeta ON mabmeta.donor_id = donormeta.donor_id
diff --git a/resources/queries/cds/alignment_mab_metadata.sql b/resources/queries/cds/alignment_mab_metadata.sql
@@ -1,4 +1,4 @@
 SELECT *
 FROM cds.alignment a
-    INNER JOIN cds.antibody_sequence aseq ON a.sequence_id = aseq.sequence_id
-    INNER JOIN cds.MAbMetadata mm ON aseq.mab_id = mm.mab_id
+    INNER JOIN cds.donor_mab_sequence dseq ON a.sequence_id = dseq.sequence_id
+    INNER JOIN cds.MAbMetadata mm ON dseq.mab_id = mm.mab_id
diff --git a/resources/queries/cds/antibody_sequence_header_source.sql b/resources/queries/cds/antibody_sequence_header_source.sql
diff --git a/resources/queries/cds/donor_mab.sql b/resources/queries/cds/donor_mab.sql
@@ -0,0 +1,4 @@
+SELECT DISTINCT dms.mab_id,
+                dms.donor_id
+FROM donor_mab_sequence AS dms
+WHERE dms.donor_id IS NOT NULL
diff --git a/resources/queries/cds/donor_mab_sequence_alignment.sql b/resources/queries/cds/donor_mab_sequence_alignment.sql
@@ -0,0 +1,3 @@
+SELECT *
+FROM alignment ali
+         JOIN donor_mab_sequence AS dms ON dms.sequence_id = ali.sequence_id
diff --git a/resources/queries/cds/donor_mab_sequence_germline.sql b/resources/queries/cds/donor_mab_sequence_germline.sql
@@ -0,0 +1,12 @@
+SELECT dms.mab_id,
+       dms.donor_id,
+       sgl.sequence_id,
+       sgl.allele,
+       sgl.percent_identity,
+       sgl.matches,
+       sgl.alignment_length,
+       sgl.score,
+       sgl.container,
+       sgl.run_application
+FROM sequence_germline AS sgl
+         JOIN donor_mab_sequence AS dms ON dms.sequence_id = sgl.sequence_id
diff --git a/resources/queries/cds/donor_mab_sequence_header_source.sql b/resources/queries/cds/donor_mab_sequence_header_source.sql
@@ -0,0 +1,15 @@
+SELECT seq.sequence_id,
+       dms.mab_id,
+       dms.donor_id,
+       mab.mab_name_std,
+       don.donor_code,
+       sqh.header,
+       src.source,
+       seq.sequence_aa,
+       seq.sequence_nt
+FROM sequence seq
+         JOIN donor_mab_sequence AS dms ON dms.sequence_id = seq.sequence_id
+         JOIN sequence_header AS sqh ON sqh.sequence_id = seq.sequence_id
+         JOIN header_source AS src ON src.source_id = sqh.source_id
+         LEFT JOIN mabMetadata AS mab ON mab.mab_id = dms.mab_id
+         LEFT JOIN donor_metadata AS don ON don.donor_id = dms.donor_id
diff --git a/resources/queries/cds/ds_mabmetasummary.sql b/resources/queries/cds/ds_mabmetasummary.sql
@@ -22,8 +22,9 @@ SELECT
   mabmeta.mab_hxb2_location,
   mabmeta.mab_ab_binding_type,
   mabmeta.mab_isotype,
-  mabmeta.mab_donor_species
+  donormeta.donor_species AS mab_donor_species
 
 FROM cds.MAbMix mix
 JOIN cds.MAbMixMetadata mixmeta ON (mixmeta.container = mix.container AND mixmeta.mab_mix_id = mix.mab_mix_id)
-JOIN cds.MAbMetadata mabmeta ON (mabmeta.container = mix.container AND mabmeta.mab_id = mix.mab_id)
+JOIN cds.MAbMetadata mabmeta ON (mabmeta.container = mix.container AND mabmeta.mab_id = mix.mab_id)
+LEFT JOIN cds.donor_metadata donormeta ON (donormeta.container = mix.container AND donormeta.donor_id = mabmeta.donor_id)
diff --git a/resources/queries/cds/lineage_alignment.sql b/resources/queries/cds/lineage_alignment.sql
diff --git a/resources/queries/cds/lineage_sequence_germline.sql b/resources/queries/cds/lineage_sequence_germline.sql
diff --git a/resources/queries/cds/sequence_germline_mab_metadata.sql b/resources/queries/cds/sequence_germline_mab_metadata.sql
@@ -1,4 +1,4 @@
 SELECT * 
 FROM cds.sequence_germline sg
-    INNER JOIN cds.antibody_sequence aseq ON sg.sequence_id = aseq.sequence_id
-    INNER JOIN cds.MAbMetadata mm ON aseq.mab_id = mm.mab_id
+    INNER JOIN cds.donor_mab_sequence dseq ON sg.sequence_id = dseq.sequence_id
+    INNER JOIN cds.MAbMetadata mm ON dseq.mab_id = mm.mab_id
diff --git a/resources/schemas/cds.xml b/resources/schemas/cds.xml
@@ -764,14 +764,13 @@
             <column columnName="mab_isotype">
                 <description>A classification indicating the type of the antibody.</description>
             </column>
-            <column columnName="mab_donorid">
+            <column columnName="donor_id">
                 <description>Unique ID assigned to the donor of the mab.</description>
-            </column>
-            <column columnName="mab_donor_species">
-                    <description>The species of the donor of the mab. The species in which the immune response was stimulated.</description>
-            </column>
-            <column columnName="mab_donor_clade">
-                <description>The clade of the virus strain (subtype) that infected the donor.</description>
+                <fk>
+                    <fkDbSchema>cds</fkDbSchema>
+                    <fkTable>donor_metadata</fkTable>
+                    <fkColumnName>donor_id</fkColumnName>
+                </fk>
             </column>
             <column columnName="mab_class_id">
                 <description>A reference ID to the `antibody_class` table, which indicates the class of the antibody.</description>
@@ -1801,6 +1800,8 @@
             <column columnName="sequence_nt">
                 <description>The sequence as nucleotides.</description>
             </column>
+            <column columnName="sequence_aa"/>
+            <column columnName="chain"/>
             <column columnName="container"/>
         </columns>
     </table>
@@ -1871,8 +1872,10 @@
             <column columnName="container"/>
         </columns>
     </table>
-    <table tableName="antibody_sequence" tableDbType="TABLE">
+    <table tableName="donor_mab_sequence" tableDbType="TABLE">
         <columns>
+            <column columnName="row_id"/>
+            <column columnName="container"/>
             <column columnName="mab_id">
                 <description>The CDS mAb ID.</description>
                 <fk>
@@ -1883,11 +1886,19 @@
             </column>
             <column columnName="sequence_id">
                 <description>The sequence ID associated with a given mAB ID and chain.</description>
+                <fk>
+                    <fkDbSchema>cds</fkDbSchema>
+                    <fkTable>sequence</fkTable>
+                    <fkColumnName>sequence_id</fkColumnName>
+                </fk>
             </column>
-            <column columnName="lineage">
-                <description>True if the sequence is a lineage sequence.</description>
+            <column columnName="donor_id">
+                <fk>
+                    <fkDbSchema>cds</fkDbSchema>
+                    <fkTable>donor_metadata</fkTable>
+                    <fkColumnName>donor_id</fkColumnName>
+                </fk>
             </column>
-            <column columnName="container"/>
         </columns>
     </table>
     <table tableName="alignment" tableDbType="TABLE">
@@ -1896,9 +1907,6 @@
             <column columnName="sequence_id">
                 <description>The CDS sequence ID.</description>
             </column>
-            <column columnName="locus">
-                <description>Gene locus (chain type). Note that this field uses a controlled vocabulary that is meant to provide a generic classification of the locus, not necessarily the correct designation according to a specific nomenclature.</description>
-            </column>
             <column columnName="stop_codon">
                 <description>True if the aligned sequence contains a stop codon.</description>
             </column>
@@ -2221,4 +2229,42 @@
             <column columnName="container"/>
         </columns>
     </table>
+    <table tableName="donor_metadata" tableDbType="TABLE">
+        <columns>
+            <column columnName="donor_id"/>
+            <column columnName="donor_lanl_id"/>
+            <column columnName="donor_code"/>
+            <column columnName="donor_species"/>
+            <column columnName="donor_clade"/>
+            <column columnName="container"/>
+        </columns>
+    </table>
+    <table tableName="pab_sequence" tableDbType="TABLE">
+        <columns>
+            <column columnName="pab_id"/>
+            <column columnName="sequence_id"/>
+            <column columnName="container"/>
+        </columns>
+    </table>
+    <table tableName="pab_sequence_study" tableDbType="TABLE">
+        <columns>
+            <column columnName="row_id"/>
+            <column columnName="prot"/>
+            <column columnName="subject_id"/>
+            <column columnName="study_day"/>
+            <column columnName="specimen_type"/>
+            <column columnName="pab_id"/>
+            <column columnName="lab_code"/>
+            <column columnName="seq_method"/>
+            <column columnName="assay_identifier"/>
+            <column columnName="container"/>
+        </columns>
+    </table>
+    <table tableName="antibody_structure" tableDbType="TABLE">
+        <columns>
+            <column columnName="structure_id"/>
+            <column columnName="mab_id"/>
+            <column columnName="container"/>
+        </columns>
+    </table>
 </tables>