From 00d3f7418cfb5d9b2de97da6edb3e22815aeadf1 Mon Sep 17 00:00:00 2001
From: Stefan Bienert <stefan.bienert@unibas.ch>
Date: Thu, 9 Nov 2023 14:31:49 +0100
Subject: [PATCH] Use reference DB info from JSON input
---
convert_to_modelcif.py | 58 +++++++++++++++++++++++++-----------------
1 file changed, 34 insertions(+), 24 deletions(-)
diff --git a/convert_to_modelcif.py b/convert_to_modelcif.py
index a2c3747..c6dcd81 100755
--- a/convert_to_modelcif.py
+++ b/convert_to_modelcif.py
@@ -364,38 +364,48 @@ def _cast_release_date(release_date):
raise
+def _cmp_ref_dbs(db_dct, db_objs):
+ """Compare a reference DB dict to a list of ReferenceDatabase objects.
+ Note: does not check the DB name!"""
+ for obj in db_objs:
+ if db_dct["release_date"] != obj.release_date:
+ continue
+ if db_dct["version"] != obj.version:
+ continue
+ for url in db_dct["location_url"]:
+ if url == obj.url:
+ return True
+
+ return False
+
+
def _get_modelcif_ref_dbs(meta_json):
"""Get sequence databases used for monomer features."""
# vendor formatting for DB names/ URLs, extend on KeyError
# ToDo: adapt to new JSON input
- sdb_dct = {} # 'sequence database list', starts as dict
+ sdb_lst = {} # 'sequence database list' starts as dict since we need to
+ # compare DBs between the different monomers.
+ i = 0
for data in meta_json.values():
+ i += 1
for db_name, vdct in data["databases"].items():
- if vdct["version"] == "NA":
- vdct["version"] = None
vdct["release_date"] = _cast_release_date(vdct["release_date"])
- # if DB already exists, check URL and version
- if db_name in sdb_dct:
- # ToDo: switch URL to the actual URL read from JSON
- if (
- sdb_dct[db_name].version != vdct["version"]
- or sdb_dct[db_name].url != vdct["location_url"][0]
- ):
- raise RuntimeError(
- "Database versions or URLs differ for "
- + f"'{db_name}': '{sdb_dct[db_name].version}/ "
- + f"{sdb_dct[db_name].url}' vs. '{vdct['version']}/ "
- + f"{vdct['location_url'][0]}'"
+ if db_name in sdb_lst:
+ if _cmp_ref_dbs(vdct, sdb_lst[db_name]):
+ continue
+ else:
+ sdb_lst[db_name] = []
+ for url in vdct["location_url"]:
+ sdb_lst[db_name].append(
+ modelcif.ReferenceDatabase(
+ db_name,
+ url,
+ version=vdct["version"],
+ release_date=vdct["release_date"],
)
- # ToDo: deal with DBs with multiple URLs
- sdb_dct[db_name] = modelcif.ReferenceDatabase(
- db_name,
- vdct["location_url"][0],
- version=vdct["version"],
- release_date=vdct["release_date"],
- )
+ )
- return sdb_dct.values()
+ return [x for sublist in sdb_lst.values() for x in sublist]
def _store_as_modelcif(
@@ -472,7 +482,7 @@ def _store_as_modelcif(
system.target_entities,
model,
sw_dct,
- # ToDo: _storte_as_modelcif should not use __meta__, __meta__ is
+ # ToDo: _store_as_modelcif should not use __meta__, __meta__ is
# tool specific
_get_modelcif_ref_dbs(data_json["__meta__"]),
)
--
GitLab