Merge pull request #498 from ssjunnebo/pod5_delivery

Exclude pod5 from delivery by default

Author: ssjunnebo

VERSIONLOG.md

@@ -1,5 +1,9 @@
 # TACA Version Log
 
+##20251121.1
+
+Exclude pod5 from delivery by default
+
 ## 20251106.1
 
 Improve logging

taca/__init__.py

@@ -1,3 +1,3 @@
 """Main TACA module"""
 
-__version__ = "1.6.12"
+__version__ = "1.6.13"

taca/organise/cli.py

@@ -13,13 +13,19 @@
     required=True,
     help="Project ID (e.g. P12345)",
 )  # future todo: option to organise all flowcells in a project
+@click.option(
+    "--include_pod5",
+    is_flag=True,
+    default=False,
+    help="Include pod5 files when organising Nanopore flowcells. Default is False.",
+)
 @click.argument("flowcells")
-def organise_flowcells(flowcells, project):
+def organise_flowcells(flowcells, project, include_pod5):
     """Organise FLOWCELLS.
 
     FLOWCELLS is the name of one or more sequencing flowcells, separated by a comma. e.g.:
     241122_VH00204_464_AAG77JJN5,241120_VH00202_453_AAG76JJM7
     """
     flowcells_to_organise = flowcells.split(",")
     for fc in flowcells_to_organise:
-        organise.organise_flowcell(fc, project)
+        organise.organise_flowcell(fc, project, include_pod5)

taca/organise/flowcells.py

@@ -11,9 +11,11 @@
 logger = logging.getLogger(__name__)
 
 
-def get_flowcell_object(flowcell, project):
+def get_flowcell_object(flowcell, project, include_pod5=False):
     if re.match(filesystem.RUN_RE_ONT, flowcell):
-        return NanoporeFlowcell(flowcell=flowcell, project_id=project)
+        return NanoporeFlowcell(
+            flowcell=flowcell, project_id=project, include_pod5=include_pod5
+        )
     elif re.match(filesystem.RUN_RE_ILLUMINA, flowcell):
         return IlluminaFlowcell(flowcell=flowcell, project_id=project)
     elif re.match(filesystem.RUN_RE_ELEMENT, flowcell):
@@ -44,13 +46,14 @@ def create_org_dir(self):
 class NanoporeFlowcell(Flowcell):
     """Defines a Nanopore Flowcell"""
 
-    def __init__(self, flowcell, project_id):
+    def __init__(self, flowcell, project_id, include_pod5=False):
         super().__init__(flowcell, project_id)
         self.destination_path = CONFIG.get("organise").get("nanopore_path")
         self.organised_project_dir = os.path.join(self.destination_path, project_id)
         self.tar_file = self.fc_id + ".tar"
         self.tar_path = os.path.join(self.organised_project_dir, self.tar_file)
         self.md5_path = self.tar_path + ".md5"
+        self.include_pod5 = include_pod5
 
     def organise_data(self):
         """Tarball data into ONT_TAR"""
@@ -59,7 +62,13 @@ def organise_data(self):
         tar_err = os.path.join(self.organised_project_dir, "tar.err")
         with filesystem.chdir(self.incoming_path):
             with open(tar_err, "w") as error_file:
-                tar_command = ["tar", "-cvf", self.tar_path, self.fc_id]
+                tar_command = [
+                    "tar",
+                    *(["--exclude=pod5*"] if not self.include_pod5 else []),
+                    "-cvf",
+                    self.tar_path,
+                    self.fc_id,
+                ]
                 result = subprocess.run(tar_command, stderr=error_file)
                 if result.returncode != 0:
                     logger.error(

taca/organise/organise.py

@@ -7,9 +7,9 @@
 logger = logging.getLogger(__name__)
 
 
-def organise_flowcell(flowcell, project):
+def organise_flowcell(flowcell, project, include_pod5=False):
     """Determine flowcell type and organise the data accordingly."""
-    flowcell_object = get_flowcell_object(flowcell, project)
+    flowcell_object = get_flowcell_object(flowcell, project, include_pod5)
     flowcell_object.create_org_dir()
     flowcell_object.organise_data()
     logger.info(f"Finished organisation of flowcell {flowcell}.")