account for neu vnb/nnb data

Author: miloswrath

dsmc.py

@@ -39,6 +39,33 @@
 SESSION_PATTERN = re.compile(r"_ses-(?P<session>[^_]+)")
 
 
+def normalize_session_label(session: str) -> str:
+    """Strip common prefixes so labels are comparable (e.g., ses-1 -> 1)."""
+    session = session.strip()
+    return session[4:] if session.startswith("ses-") else session
+
+
+def load_offloaded_pairs(path: Path) -> Set[Tuple[str, str, str]]:
+    """
+    Read offloaded task records.
+
+    Returns a set of (task, subject, session) tuples using normalized session labels.
+    """
+    pairs: Set[Tuple[str, str, str]] = set()
+    if not path.exists():
+        return pairs
+
+    with path.open(newline="") as f:
+        reader = csv.DictReader(f)
+        for row in reader:
+            task = row.get("task", "").strip().upper()
+            subject = row.get("sub", "").strip()
+            session = normalize_session_label(row.get("session", ""))
+            if task and subject and session:
+                pairs.add((task, subject, session))
+    return pairs
+
+
 def parse_args() -> argparse.Namespace:
     parser = argparse.ArgumentParser(
         description="Count complete subjects per session and list missing tasks."
@@ -109,7 +136,10 @@ def collect_subject_sessions(subject_dir: Path) -> Dict[str, Set[str]]:
 
 
 def gather_dataset_stats(
-    dataset_root: Path, dataset_label: str, expected_tasks: Set[str]
+    dataset_root: Path,
+    dataset_label: str,
+    expected_tasks: Set[str],
+    offloaded_pairs: Set[Tuple[str, str, str]],
 ) -> Tuple[Dict[str, int], List[Tuple[str, str, str, str]], Set[str]]:
     """
     Compute completeness counts and missing task rows for one dataset.
@@ -128,6 +158,22 @@ def gather_dataset_stats(
         for session, tasks_present in session_tasks.items():
             sessions_seen.add(session)
             missing = expected_tasks - tasks_present
+
+            # Exempt VNB/NNB sessions that were run offline (recorded in offloaded.csv).
+            missing = {
+                task
+                for task in missing
+                if not (
+                    task.upper() in {"VNB", "NNB"}
+                    and (
+                        task.upper(),
+                        subject_dir.name,
+                        normalize_session_label(session),
+                    )
+                    in offloaded_pairs
+                )
+            }
+
             if not missing:
                 complete_counts[session] += 1
             else:
@@ -170,6 +216,10 @@ def main() -> None:
     output_dir: Path = args.output_dir
     output_dir.mkdir(parents=True, exist_ok=True)
 
+    offloaded_pairs = load_offloaded_pairs(
+        Path(__file__).resolve().parent / "offloaded.csv"
+    )
+
     # If no explicit tasks were provided, include any additional tasks found on disk.
     if not args.tasks:
         expected_tasks = discover_tasks(data_root) or DEFAULT_TASKS
@@ -183,7 +233,7 @@ def main() -> None:
             continue
 
         counts, missing, sessions_seen = gather_dataset_stats(
-            dataset_path, dataset_label, expected_tasks
+            dataset_path, dataset_label, expected_tasks, offloaded_pairs
         )
 
         for session in sorted(sessions_seen, key=session_sort_key):

meta/dsmc_complete_counts.csv

@@ -2,5 +2,4 @@ dataset,session,complete_subjects
 int,ses-1,68
 int,ses-2,46
 int,ses-3,24
-int,ses-33,0
 obs,ses-1,214

meta/dsmc_missing_tasks.csv

@@ -1,101 +1,22 @@
 dataset,site,subject,session,missing_task
-int,NE,0,ses-33,DSST
-int,NE,0,ses-33,DWL
-int,NE,0,ses-33,FN
-int,NE,0,ses-33,LC
-int,NE,0,ses-33,NF
-int,NE,0,ses-33,PC
-int,NE,0,ses-33,SM
-int,NE,0,ses-33,VNB
-int,NE,0,ses-1,ATS
-int,NE,0,ses-1,DSST
-int,NE,0,ses-1,FN
-int,NE,0,ses-1,LC
-int,NE,0,ses-1,NF
-int,NE,0,ses-1,NNB
-int,NE,0,ses-1,NTS
-int,NE,0,ses-1,SM
-int,NE,0,ses-1,VNB
-int,NE,00,ses-33,AF
-int,NE,00,ses-33,ATS
-int,NE,00,ses-33,DSST
-int,NE,00,ses-33,DWL
-int,NE,00,ses-33,FN
-int,NE,00,ses-33,LC
-int,NE,00,ses-33,NNB
-int,NE,00,ses-33,NTS
-int,NE,00,ses-33,PC
-int,NE,00,ses-33,SM
-int,NE,00,ses-33,VNB
-int,NE,00,ses-33,WL
-int,NE,00,ses-1,AF
-int,NE,00,ses-1,ATS
-int,NE,00,ses-1,DSST
-int,NE,00,ses-1,DWL
-int,NE,00,ses-1,FN
-int,NE,00,ses-1,LC
-int,NE,00,ses-1,NNB
-int,NE,00,ses-1,NTS
-int,NE,00,ses-1,PC
-int,NE,00,ses-1,SM
-int,NE,00,ses-1,VNB
-int,NE,00,ses-1,WL
-int,NE,1,ses-1,AF
-int,NE,1,ses-1,ATS
-int,NE,1,ses-1,DSST
-int,NE,1,ses-1,FN
-int,NE,1,ses-1,LC
-int,NE,1,ses-1,NF
-int,NE,1,ses-1,NNB
-int,NE,1,ses-1,NTS
-int,NE,1,ses-1,SM
-int,NE,1,ses-1,VNB
-int,NE,1,ses-1,WL
-int,NE,1111,ses-1,AF
-int,NE,1111,ses-1,ATS
-int,NE,1111,ses-1,DSST
-int,NE,1111,ses-1,DWL
-int,NE,1111,ses-1,FN
-int,NE,1111,ses-1,NNB
-int,NE,1111,ses-1,NTS
-int,NE,1111,ses-1,PC
-int,NE,1111,ses-1,SM
-int,NE,1111,ses-1,VNB
-int,NE,9005,ses-3,VNB
 int,NE,9007,ses-1,WL
 int,NE,9008,ses-1,SM
 int,NE,9009,ses-2,DSST
 int,NE,9009,ses-2,VNB
 int,NE,9010,ses-3,NF
 int,NE,9010,ses-3,NTS
-int,NE,9013,ses-3,NNB
-int,NE,9013,ses-2,VNB
-int,NE,9015,ses-2,VNB
 int,NE,9017,ses-3,LC
 int,NE,9017,ses-3,VNB
-int,NE,9019,ses-3,VNB
 int,NE,9020,ses-1,NF
 int,NE,9021,ses-2,VNB
-int,NE,9021,ses-1,VNB
-int,NE,9023,ses-1,VNB
-int,NE,9024,ses-1,VNB
-int,NE,9028,ses-2,NNB
 int,NE,9028,ses-2,VNB
 int,NE,9028,ses-1,VNB
 int,NE,9029,ses-1,NNB
 int,NE,9029,ses-1,VNB
 int,NE,9031,ses-2,VNB
 int,NE,9033,ses-1,DWL
-int,NE,9033,ses-1,NNB
 int,NE,9033,ses-1,NTS
-int,NE,9033,ses-1,VNB
-int,NE,9034,ses-1,VNB
-int,NE,9035,ses-1,NNB
-int,NE,9035,ses-1,VNB
-int,NE,9038,ses-1,VNB
 int,NE,9045,ses-1,DSST
-int,NE,9999,ses-1,AF
-int,NE,9999,ses-1,SM
 int,UI,8004,ses-3,AF
 int,UI,8005,ses-2,NNB
 int,UI,8005,ses-2,NTS

offloaded.csv

@@ -0,0 +1,18 @@
+task,sub,counterbalance,session
+VNB,9013,C,ses-2
+VNB,9021,B,ses-1
+VNB,9023,C,ses-1
+VNB,9024,C,ses-1
+VNB,9005,B,ses-3
+VNB,9015,C,ses-2
+VNB,9034,C,ses-1
+VNB,9033,B,ses-1
+VNB,9019,C,ses-3
+VNB,9038,C,ses-1
+VNB,9035,C,ses-1
+VNB,9031,B,ses-3
+NNB,9013,B,ses-3
+NNB,9033,B,ses-1
+NNB,9035,C,ses-1
+NNB,9028,B,ses-2
+