pubmed needed

Load/gene_alleles.py

@@ -483,6 +483,49 @@ def create_alpha_alleles(cursor, org_dict, feature_id, cvterm_id, db_id, pub_id)
         )
 
 
+def add_gene_GA(cursor, organism_id, feature_id, cvterm_id, dbxref_id, pub_id, db_id):
+    """Add data to test G24 banc and bangd operations.
+
+    Args:
+        cursor: <sql connection cursor> connection to testdb
+        organism_id: <int> organism id
+        feature_id: <dict> feature name to id
+        cvterm_id: <dict> cvterm name to id
+        dbxref_id: <dict> dbxref name to dbxref id
+        pub_id: <int> id of pub
+        db_id: <dict> db name to db id
+
+    """
+    create_gene_alleles(
+        cursor, organism_id, feature_id, cvterm_id, db_id, pub_id,
+        num_genes=10,
+        num_alleles=1,
+        gene_prefix='GAproptest',
+        allele_prefix=None,
+        tool_prefix='Cjk',
+        pub_format="GAproptest_title_"
+        )
+    # G24 data. feature cvterm props
+    # fc_sql = """ INSERT INTO feature_cvterm (feature_id, cvterm_id, pub_id) VALUES (%s, %s, %s) RETURNING feature_cvterm_id """
+    # fcp_sql = """ INSERT INTO feature_cvtermprop (feature_cvterm_id, type_id, value, rank) VALUES (%s, %s, %s, %s) """
+    data = [
+        {'cvterm': 'date', 'cvname': 'feature_cvtermprop type', 'value': '19671008'},
+        {'cvterm': 'provenance', 'cvname': 'FlyBase miscellaneous CV', 'value': 'FlyBase'},
+        {'cvterm': 'evidence_code', 'cvname': 'FlyBase miscellaneous CV', 'value': 'inferred from direct assay'},
+        {'cvterm': 'located_in', 'cvname': 'relationship', 'value': None}]
+
+    for i in range(10):
+        # create feature cvterm
+        cursor.execute(fc_sql,(feature_id['GAproptest{}[Cjk1]'.format(i + 1)], cvterm_id['in vitro construct'], pub_id))
+        iv_id = cursor.fetchone()[0]
+        cursor.execute(fc_sql, (feature_id['GAproptest{}[Cjk1]'.format(i+1)], cvterm_id['amorphic allele - molecular evidence'], pub_id))
+        iv_id = cursor.fetchone()[0]
+
+        # create feature cvterm props
+        for item in data:
+            cursor.execute(fcp_sql, (fc_id, cvterm_id[item['cvterm']], item['value'], 0))
+
+
 def add_gene_G24(cursor, organism_id, feature_id, cvterm_id, dbxref_id, pub_id, db_id):
     """Add data to test G24 banc and bangd operations.
 

Load/pubs.py

@@ -5,12 +5,15 @@ def add_pub_data(cursor, feature_id, cv_id, cvterm_id, db_id, db_dbxref):
     pub_sql = """ INSERT INTO pub (title, type_id, uniquename, pyear, miniref) VALUES (%s, %s, %s, %s, %s) RETURNING pub_id """
     pubprop_sql = """ INSERT INTO pubprop (pub_id, rank, type_id, value) VALUES (%s, %s, %s, %s) """
     editor_sql = """ INSERT INTO pubauthor (pub_id, rank, surname, givennames, editor) VALUES (%s, %s, %s, %s, %s) """
+    pub_dbxref_sql = """ INSERT INTO pub_dbxref (pub_id, dbxref_id) VALUES (%s, %s) """
 
     for i in range(2, 9):
         cursor.execute(pub_sql, ('Nature_{}'.format(i), cvterm_id['computer file'], 'FBrf000000{}'.format(i), '1967', 'miniref_{}'.format(i)))
         pub_id = cursor.fetchone()[0]
         feature_id['Nature_{}'.format(i)] = pub_id
         print("Pub'{}' id = {}".format('Nature_{}'.format(i), pub_id))
+        cursor.execute(pub_dbxref_sql, (pub_id, db_dbxref['pubmed'][f'{i}']))
+
     cursor.execute(pub_sql, ('unattributed', cvterm_id['unattributed'], 'unattributed', '1973', 'miniref_10'))
     pub_id = cursor.fetchone()[0]
     feature_id['unattributed'] = pub_id
@@ -48,7 +51,6 @@ def add_pub_data(cursor, feature_id, cv_id, cvterm_id, db_id, db_dbxref):
     parent_pub_id = cursor.fetchone()[0]
 
     # create general multipubs for testing
-    pub_dbxref_sql = """ INSERT INTO pub_dbxref (pub_id, dbxref_id) VALUES (%s, %s) """
     for i in range(4, 14):
         cursor.execute(pub_sql, ('Journal_{}'.format(i+1), cvterm_id['journal'], 'multipub:temp_{}'.format(i), '2018', 'miniref_{}'.format(i+1)))
         pub_id = cursor.fetchone()[0]

data/cv_cvterm.yaml

@@ -54,11 +54,11 @@ FlyBase miscellaneous CV: [
     'environ1', 'environ2', 'environ3','environ4', 'environ5', 'faint', 'functional group', 'female',
     'in vitro construct',
     'inferred from direct assay', 'inferred from mutant phenotype', 'isolated cells', 'natural population',
-    'male', 'misc 1', 'misc 2', 'misc 3', 'multi-individual sample',
+    'male', 'misc 1', 'misc 2', 'misc 3', 'multi-individual sample', 'partially',
     'pheno1', 'pheno2', 'pheno3', 'pheno4', 'pheno5',
     'photoactivatable fluorescent protein', 'protein detection tool', 'project',
     'qualifier', 'reagent collection', 'RNA detection tool', 'single balancer', 'spontaneous',
-    'split system combination',
+    'split system combination', 'suppressible',
     'transcriptome', 'umbrella project', 'unspecified']
 
 GenBank feature qualifier: [

data/db_dbxref.yaml

@@ -7,7 +7,7 @@ FBdv: []
 GB: ['GB1', 'GB2']
 GO: []
 DOI: []
-pubmed: []
+pubmed: ['1','2','3','4','5','6','7','8','9','10','11','12','13','14']
 isbn: []
 issn: ['1111-1111', '2222-2222', '3333-3333', '4444-4444']
 PMCID: []