fix(IPVC-2399): fix gene id back fill migration script, update paths … (

#29)
invitae · Apr 26, 2024 · b915f31 · b915f31
1 parent 8ff7464
commit b915f31
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 16 deletions.
diff --git a/misc/gene-update/backfill_gene_id.py b/misc/gene-update/backfill_gene_id.py
@@ -7,6 +7,7 @@
 
 import uta
 from uta.models import Gene, Transcript
+from uta.tools.file_utils import open_file
 
 
 logger = None
@@ -22,7 +23,7 @@ def backfill_gene(uta_session: Session, gene_update_file: str) -> None:
     now_ts = datetime.now()
     i = 0
     new_genes = []
-    with open(gene_update_file) as f:
+    with open_file(gene_update_file) as f:
         for line in f:
             if line.startswith("gene_id"):
                 continue
@@ -70,7 +71,7 @@ def backfill_transcript(uta_session: Session, transcript_update_file: str) -> No
     tx_ac_to_gene_id = {}
 
     logger.info(f"Reading transcript to gene id mappings from {transcript_update_file}")
-    with open(transcript_update_file) as f:
+    with open_file(transcript_update_file) as f:
         for line in f:
             if line.startswith("origin"):
                 continue

diff --git a/misc/gene-update/upgrade-uta-schema.sh b/misc/gene-update/upgrade-uta-schema.sh
@@ -44,8 +44,8 @@ alembic -c etc/alembic.ini upgrade 595a586e6de7
 # run data migration to back fill gene_id
 python misc/gene-update/backfill_gene_id.py \
   postgresql://uta_admin:@localhost/uta \
-  /workdir/backfill/gene_update.tsv \
-  /workdir/backfill/transcript_update.tsv
+  misc/gene-update/gene_update.tsv.gz \
+  misc/gene-update/transcript_update.tsv.gz
 
 # run Alembic migrations to add constraints and update existing views
 alembic -c etc/alembic.ini upgrade head

diff --git a/src/alembic/versions/f85dd97bd9f5_set_gene_id_and_primary_and_foreign_keys.py b/src/alembic/versions/f85dd97bd9f5_set_gene_id_and_primary_and_foreign_keys.py
@@ -35,23 +35,17 @@ def upgrade() -> None:
         unique=False,
         schema="uta",
     )
-    op.create_foreign_key(
-        "fk_uta_transcript_gene_gene_id",
-        "transcript",
-        "gene",
-        ["gene_id"],
-        ["gene_id"],
-        source_schema="uta",
-        referent_schema="uta",
-        onupdate="RESTRICT",
-        ondelete="RESTRICT",
-    )
+    op.create_foreign_key(None, 'transcript', 'gene', ['gene_id'], ['gene_id'], source_schema='uta', referent_schema='uta')
     # ### end Alembic commands ###
 
     # ### handle first part of hgnc -> gene_symbol column rename ###
     op.add_column("gene", sa.Column("symbol", sa.Text(), nullable=True), schema="uta")
     op.create_index(op.f("ix_uta_gene_symbol"), "gene", ["symbol"], unique=False, schema="uta")
     op.execute("UPDATE gene SET symbol = hgnc;")
+    op.alter_column('gene', 'symbol',
+                    existing_type=sa.TEXT(),
+                    nullable=False,
+                    schema='uta')
     # ### end of hgnc -> gene_symbol column rename ###
 
     # ### updates required to existing views needed to drop hgnc from transcript. ###
@@ -301,7 +295,7 @@ def downgrade() -> None:
     # ### end of updates to views ###
 
     # ### commands auto generated by Alembic - please adjust! ###
-    op.drop_constraint("fk_uta_transcript_gene_gene_id", "transcript", schema="uta", type_="foreignkey")
+    op.drop_constraint(None, 'transcript', schema='uta', type_='foreignkey')
     op.drop_index(op.f("ix_uta_transcript_gene_id"), table_name="transcript", schema="uta")
     op.alter_column("transcript", "gene_id",
                existing_type=sa.TEXT(),