fixed epic+ qualityMask

FoxoTech · May 18, 2021 · d077c4e · d077c4e
1 parent 264e3e8
commit d077c4e
Show file tree

Hide file tree

Showing 12 changed files with 42 additions and 64,158 deletions.
diff --git a/docs/example_data/epic_plus/sample_sheet_meta_data.pkl b/docs/example_data/epic_plus/sample_sheet_meta_data.pkl
diff --git a/docs/release-history.md b/docs/release-history.md
@@ -1,5 +1,13 @@
 # Release History
 
+## v1.4.5
+- fixed qualityMask for epic+
+
+## v1.4.4
+- faster circleci testing
+- mouse probes have duplicate names, breaking dye-bias step, so it will fallback to linear-dye when duplicates are present
+- added more mouse array test coverage
+
 ## v1.4.0
 - now uses sesame's infer_type_I_channel function to detect and correct probe switching, if sesame=True
 - uses sesame's nonlinear dye bias correction function, if sesame=True

diff --git a/methylprep/download/array_express.py b/methylprep/download/array_express.py
@@ -30,12 +30,12 @@ def ae_download(ae_id, series_path, ae_platforms, clean=True):
     series_dir = Path(series_path)
     sdrf_filename = f"{ae_id}.sdrf.txt"
 
-    if not os.path.exists(series_path):
+    if not Path(series_path).exists():
         raise FileNotFoundError(f'{ae_id} directory not found')
 
     for platform in ae_platforms:
-        if not os.path.exists(f"{series_path}/{platform}"):
-            os.mkdir(f"{series_path}/{platform}")
+        if not Path(f"{series_path}/{platform}").exists():
+            Path(f"{series_path}/{platform}").mkdir(parents=True, exist_ok=True)
 
     ftp = FTP('ftp.ebi.ac.uk', timeout=59)
     ftp.login()

diff --git a/methylprep/download/geo.py b/methylprep/download/geo.py
@@ -83,7 +83,7 @@ def geo_download(geo_id, series_path, geo_platforms, clean=True, decompress=True
 
     for platform in geo_platforms:
         if not Path(f"{series_path}/{platform}").exists():
-            Path(f"{series_path}/{platform}").mkdir()
+            Path(f"{series_path}/{platform}").mkdir(parents=True, exist_ok=True)
 
     ftp = FTP('ftp.ncbi.nlm.nih.gov', timeout=120) # 2 mins
     ftp.login()
@@ -362,7 +362,7 @@ def pipeline_find_betas_any_source(**kwargs):
         working = tempfile.TemporaryDirectory(dir=EFS)
     elif not kwargs.get('efs') and kwargs.get('data_dir') != None:
         if not Path(kwargs.get('data_dir')).exists():
-            Path(kwargs.get('data_dir')).mkdir()
+            Path(kwargs.get('data_dir')).mkdir(parents=True, exist_ok=True)
         working = tempfile.TemporaryDirectory(dir=kwargs['data_dir'])
         EFS = working.name
     elif not kwargs.get('efs') and kwargs.get('data_dir') == None:

diff --git a/methylprep/download/miniml.py b/methylprep/download/miniml.py
@@ -167,7 +167,7 @@ def download_miniml(geo_id, series_path, remove_tgz=False, verbose=False):
     expected_miniml_filename = f"{geo_id}_family.xml"
 
     if not Path(series_path).exists():
-        Path(series_path).mkdir()
+        Path(series_path).mkdir(parents=True, exist_ok=True)
 
     ftp = FTP('ftp.ncbi.nlm.nih.gov', timeout=120) # 2 mins
     ftp.login()

diff --git a/methylprep/download/process_data.py b/methylprep/download/process_data.py
@@ -169,7 +169,7 @@ def run_series_list(list_file, path, dict_only=False, batch_size=BATCH_SIZE, **k
         series_id = series_id.strip()
         series_path = Path(path, series_id) # run_series and geo_download get confused if idats already present, so this avoids that confusion
         try:
-            series_path.mkdir()
+            series_path.mkdir(parents=True, exist_ok=True)
             LOGGER.info(f"Running {series_id}")
             run_series(series_id, series_path, dict_only=dict_only, batch_size=batch_size, **kwargs)
         except (ValueError, FileNotFoundError) as e: