Make db optional

fusion_report.log.2024-06-25

@@ -0,0 +1,64 @@
+2024-06-25 09:55:25,106 - CRITICAL - fusion_report.common.net - table "mbca" already exists
+Traceback (most recent call last):
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/common/db.py", line 61, in setup
+    self.create_database()
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/common/db.py", line 84, in create_database
+    self.connection.executescript(schema.read().lower())
+sqlite3.OperationalError: table "mbca" already exists
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/app.py", line 72, in run
+    Download(params)
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/download.py", line 23, in __init__
+    self.download_all(params)
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/download.py", line 42, in download_all
+    Net.get_mitelman(self, return_err)
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/common/net.py", line 203, in get_mitelman
+    db.setup(files, delimiter="\t", skip_header=False, encoding="ISO-8859-1")
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/common/db.py", line 79, in setup
+    raise DbException(ex)
+fusion_report.common.exceptions.db.DbException: table "mbca" already exists
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/EGG-INFO/scripts/fusion_report", line 13, in <module>
+    app.run()
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/app.py", line 79, in run
+    raise AppException(ex)
+fusion_report.common.exceptions.app.AppException: table "mbca" already exists
+2024-06-25 09:56:32,905 - CRITICAL - fusion_report.common.net - unable to open database file
+Traceback (most recent call last):
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/common/db.py", line 37, in connect
+    connection = sqlite3.connect(os.path.join(path, database))
+sqlite3.OperationalError: unable to open database file
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/app.py", line 60, in run
+    self.preprocess(params)
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/app.py", line 84, in preprocess
+    self.enrich(params.db_path)
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/app.py", line 124, in enrich
+    CosmicDB(path).name: CosmicDB(path).get_all_fusions(),
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/common/singleton.py", line 12, in __call__
+    cls._instances[cls] = super(Singleton, cls).__call__(*args, **kwargs)
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/data/cosmic.py", line 15, in __init__
+    super().__init__(path, Settings.COSMIC["NAME"], Settings.COSMIC["SCHEMA"])
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/common/db.py", line 25, in __init__
+    self.connection = self.connect(path, self.database)
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/common/db.py", line 41, in connect
+    raise DbException(ex)
+fusion_report.common.exceptions.db.DbException: unable to open database file
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/EGG-INFO/scripts/fusion_report", line 13, in <module>
+    app.run()
+  File "/Users/annick.renevey/anaconda3/envs/fusion-report/lib/python3.9/site-packages/fusion_report-2.1.5-py3.9.egg/fusion_report/app.py", line 79, in run
+    raise AppException(ex)
+fusion_report.common.exceptions.app.AppException: unable to open database file

fusion_report/app.py

@@ -8,7 +8,7 @@
 from collections import defaultdict
 from typing import Any, Dict, List
 
-import rapidjson
+import json
 
 from tqdm import tqdm
 
@@ -138,7 +138,7 @@ def export_results(self, path: str, extension: str) -> None:
         if extension == "json":
             with open(dest, "w", encoding="utf-8") as output:
                 results = [fusion.json_serialize() for fusion in self.manager.fusions]
-                output.write(rapidjson.dumps(results))
+                output.write(json.dumps(results))
         elif extension == "csv":
             with open(dest, "w", encoding="utf-8") as output:
                 csv_writer = csv.writer(
@@ -248,4 +248,4 @@ def generate_multiqc(
 
         dest = f"{os.path.join(path, 'fusion_genes_mqc.json')}"
         with open(dest, "w", encoding="utf-8") as output:
-            output.write(rapidjson.dumps(configuration))
+            output.write(json.dumps(configuration))

fusion_report/args_builder.py

@@ -4,7 +4,7 @@
 from argparse import ArgumentParser, Namespace, _SubParsersAction
 from typing import Any, Dict
 
-import rapidjson
+import json
 
 from fusion_report.settings import Settings
 
@@ -21,7 +21,7 @@ class ArgsBuilder:
     def __init__(self):
         configuration = os.path.join(Settings.ROOT_DIR, "arguments.json")
         with open(configuration, "r") as config_file:
-            self.arguments: Dict[str, Any] = rapidjson.loads(config_file.read())
+            self.arguments: Dict[str, Any] = json.load(config_file)
         self.arguments["weight"] = float(100 / len(self.supported_tools))
         self.parser = ArgumentParser(
             description="""Tool for generating friendly UI custom report."""
@@ -108,6 +108,13 @@ def download_args(self, args: Dict[str, Any]) -> None:
         for mandatory in args["mandatory"]:
             download_parser.add_argument(mandatory["key"], help=mandatory["help"], type=str)
 
+        for optional in args["optionals"]:
+            download_parser.add_argument(
+                optional["key"],
+                help=optional["help"],
+                action=optional.get("action", "store"),
+            )
+
         self._cosmic(args, download_parser)
 
     def sync_args(self, args: Dict[str, Any]) -> None:

fusion_report/arguments.json

@@ -79,6 +79,23 @@
                     "help": "Output directory"
                 }
             ],
+            "optionals": [
+                {
+                    "key": "--no-cosmic",
+                    "help": "Do not download cosmic fusion database",
+                    "action": "store_true"
+                },
+                {
+                    "key": "--no-fusiongdb2",
+                    "help": "Do not download fusiongdb2 fusion database",
+                    "action": "store_true"
+                },
+                {
+                    "key": "--no-mitelman",
+                    "help": "Do not download mitelman fusion database",
+                    "action": "store_true"
+                }
+            ],
             "cosmic": [
                 {
                     "key": "--cosmic_usr",

fusion_report/common/net.py

@@ -26,6 +26,7 @@
 class Net:
     @staticmethod
     def get_cosmic_token(params: Namespace):
+        print(params.no_cosmic)
         if params.cosmic_token is not None:
             return params.cosmic_token
 

fusion_report/download.py

@@ -19,7 +19,6 @@ class Download:
     """
 
     def __init__(self, params: Namespace):
-        self.validate(params)
         self.download_all(params)
 
     def validate(self, params: Namespace) -> None:
@@ -29,28 +28,34 @@ def validate(self, params: Namespace) -> None:
         else:
             self.cosmic_token = Net.get_cosmic_token(params)
 
+
+    def download_all(self, params: Namespace) -> None:
+
         # making sure output directory exists
         if not os.path.exists(params.output):
             os.makedirs(params.output, 0o755)
 
-    def download_all(self, params: Namespace) -> None:
         """Download all databases."""
         return_err: List[str] = []
         os.chdir(params.output)
 
-        # MITELMAN
-        Net.get_mitelman(self, return_err)
+        if not params.no_mitelman:
+            # MITELMAN
+            Net.get_mitelman(self, return_err)
 
-        # FusionGDB2
-        Net.get_fusiongdb2(self, return_err)
+        if not params.no_fusiongdb2:
+            # FusionGDB2
+            Net.get_fusiongdb2(self, return_err)
 
-        # COSMIC
-        if params.qiagen:
-            Logger(__name__).info("Downloading resources from QIAGEN...")
-            Net.get_cosmic_from_qiagen(self.cosmic_token, return_err, params.output)
-        else:
-            Logger(__name__).info("Downloading resources from SANGER...")
-            Net.get_cosmic_from_sanger(self.cosmic_token, return_err)
+        if not params.no_cosmic:
+            # COSMIC
+            self.validate(params)
+            if params.qiagen:
+                Logger(__name__).info("Downloading resources from QIAGEN...")
+                Net.get_cosmic_from_qiagen(self.cosmic_token, return_err, params.output)
+            else:
+                Logger(__name__).info("Downloading resources from SANGER...")
+                Net.get_cosmic_from_sanger(self.cosmic_token, return_err)
 
         if len(return_err) > 0:
             raise DownloadException(return_err)

requirements.txt

@@ -4,7 +4,6 @@ pyyaml>=4.2b1
 Jinja2>=2.10
 MarkupSafe>=2.1.1
 pandas
-python-rapidjson
 requests
 openpyxl
 xlrd >= 2.0.0

setup.cfg

@@ -6,4 +6,4 @@ lines_after_imports=2
 lines_between_types=1
 multi_line_output=7
 use_parentheses=true
-known_third_party=colorlog,jinja2,rapidjson,setuptools,tqdm,yaml
+known_third_party=colorlog,jinja2,setuptools,tqdm,yaml