Added docker and docker compose

Dockerfile

@@ -0,0 +1,16 @@
+FROM bitnami/python:3.10
+
+RUN mkdir /pseudo-app
+
+WORKDIR /pseudo-app
+
+ADD requirements.txt .
+ADD run_pseudonymization_pipeline.py .
+ADD pseudonymization/ pseudonymization/
+ADD tests/ tests/
+
+RUN pip install -r requirements.txt
+
+USER 1001
+
+CMD ["pytest", "tests"]

compose.yml

@@ -0,0 +1,18 @@
+version: '3.0'
+services:
+  run:
+    build: .
+    volumes:
+      - /home/houfek/Work/MMCI/sequencing_pipeline/data-catalogue-playground/muni-sc/:/sc
+      - /home/houfek/Work/MMCI/sequencing_pipeline/data-catalogue-playground/seq/TRANSFER/:/TRANSFER
+      - /home/houfek/Work/MMCI/sequencing_pipeline/data-catalogue-playground/pseudonymisation_table/:/pseudonymization_tables/
+      - /home/houfek/Work/MMCI/sequencing_pipeline/data-catalogue-playground/Libraries/:/libraries
+    extra_hosts:
+      - "host.docker.internal:host-gateway"
+    command: bash -c "python run_pseudonymization_pipeline.py 
+                              /TRANSFER
+                              /sc/MiSEQ
+                              /pseudonymization_tables
+                              /libraries
+                              /sc/Libraries
+                              'MiSEQ'"

pseudonymization/helpers/config.cfg

@@ -1,3 +1,3 @@
 [miseq-config]
-PSEUDONYMIZATION-API = http://localhost:8088/api
-EXPORT-API = http://localhost:8080/api
+PSEUDONYMIZATION-API = http://host.docker.internal:8088/api
+EXPORT-API = http://host.docker.internal:8080/api

pseudonymization/process/pseudonimizer.py

@@ -25,7 +25,7 @@ def __call__(self):
         pred_pseudo_tuples = self._get_all_predictive_numbers_pseudonymize_sample_sheet()
         for pred, pseudo in pred_pseudo_tuples:
             self._pseudonymize_file_names_recursively(pred, pseudo, self.run_path)
-            self._try_pseudonimize_content_of_files(pred, pseudo)
+            self._try_pseudonimize_content_of_files(pred, pseudo) # This needs to run after the _pseudo_files_names_req
             clinical_data = ClinicalInfoFinder(self.run_path).collect_clinical_data(pred)
 
             if clinical_data:

run_pseudonymization_pipeline.py

@@ -6,15 +6,17 @@
 from pseudonymization.process.pseudonimizer import RunPseudonimizer
 
 
-def pseudonymize_miseq_runs(source_folder, destination_folder, pseudo_samples_folder, sequencing_libraries):
+def pseudonymize_miseq_runs(source_folder, destination_folder, pseudo_samples_folder, sequencing_libraries, sc_libraries):
+    shutil.copytree(sequencing_libraries, sc_libraries, dirs_exist_ok=True)
     for run in os.listdir(source_folder):
         run_path = os.path.join(source_folder, run)
         remove_miseq_run_files(run_path)
         RunPseudonimizer(run_path, pseudo_samples_folder)()
         mv_if_source_not_exist(run_path, os.path.join(destination_folder, run))
 
 
-def pseudonymize_nextseq_runs(source_folder, destination_folder, pseudo_samples_folder, sequencing_libraries):
+def pseudonymize_nextseq_runs(source_folder, destination_folder, pseudo_samples_folder, sequencing_libraries, sc_libraries):
+    shutil.copytree(sequencing_libraries, sc_libraries, dirs_exist_ok=True)
     for run in os.listdir(source_folder):
         run_path = os.path.join(source_folder, run)
         remove_nextseq_run_files(run_path)
@@ -35,6 +37,11 @@ def pseudonymize_nextseq_runs(source_folder, destination_folder, pseudo_samples_
     args = parser.parse_args()
 
     if args.run_type == "MiSEQ":
-        pseudonymize_miseq_runs(args.source_folder, args.destination_folder, args.pseudo_tables_folder, args.sequencing_libraries)
+        pseudonymize_miseq_runs(args.source_folder,
+                                args.destination_folder,
+                                args.pseudo_tables_folder,
+                                args.sequencing_libraries,
+                                args.sequencing_libraries_sc)
+        print("DONE")
     elif args.run_type == "NextSeq":
-        pseudonymize_nextseq_runs(args.source_folder, args.destination_folder, args.pseudo_tables_folder, args.sequencing_libraries)
+        pseudonymize_nextseq_runs(args.source_folder, args.destination_folder, args.pseudo_tables_folder, args.sequencing_libraries, args.sequencing_libraries_sc)

tests/test_pseudonimization.py

@@ -171,12 +171,11 @@ def test_save_clinical_data(mocker,
 
     pat = Patient("mmci_patient_12345678-1234-5678-1234-567812345621", "Fri, 1 Mar 1990 00:00:00 GMT", 1, samples)
 
-    folder_for_clinical_data = os.path.join(FAKE_RUN_FOLDER_FOR_COPY, "catalog_info_per_pred_number")
+    folder_for_clinical_data = os.path.join(FAKE_RUN_FOR_TESTING, "catalog_info_per_pred_number")
     pseudo_pred_number = "mmci_patient_12345678-1234-5678-1234-567812345688"
 
     pseudonimizer = RunPseudonimizer(FAKE_RUN_FOR_TESTING, PSEUDONYMIZATION_FILES_FOLDER)
     pseudonimizer._save_clinical_data(pat, folder_for_clinical_data, pseudo_pred_number)
 
 
-    assert os.path.exists(os.path.join(FAKE_RUN_FOLDER_FOR_COPY, "catalog_info_per_pred_number", f"{pseudo_pred_number}.json"))
-
+    assert os.path.exists(os.path.join(FAKE_RUN_FOR_TESTING, "catalog_info_per_pred_number", f"{pseudo_pred_number}.json"))