#1007 #1071 #1050 patch incorrect structure assignments

Author: NileGraddis

MANIFEST.in

@@ -1,2 +1,2 @@
-recursive-include allensdk *.py *.conf *.json *.txt *.yaml
+recursive-include allensdk *.py *.conf *.json *.txt *.yaml *.csv
 include README.md LICENSE.txt Makefile *requirements.txt

allensdk/brain_observatory/ecephys/ecephys_project_api/ecephys_project_warehouse_api.py

@@ -9,6 +9,9 @@
 from .ecephys_project_api import EcephysProjectApi
 from .utilities import rma_macros, build_and_execute
 
+from allensdk.brain_observatory.ecephys.ecephys_project_api.warehouse_patches import replace_bad_structure_assignments
+
+
 class EcephysProjectWarehouseApi(EcephysProjectApi):
 
     movie_re = re.compile(r".*natural_movie_(?P<num>\d+).npy")
@@ -205,6 +208,7 @@ def get_channels(self, channel_ids=None, probe_ids=None):
         )
 
         response.set_index("id", inplace=True)
+        replace_bad_structure_assignments(response, inplace=True)
 
         return response
 
@@ -304,4 +308,4 @@ def get_unit_analysis_metrics(self, unit_ids=None, ecephys_session_ids=None, ses
     def default(cls, **rma_kwargs):
         _rma_kwargs = {"scheme": "http", "host": "api.brain-map.org"}
         _rma_kwargs.update(rma_kwargs)
-        return cls(RmaEngine(**_rma_kwargs))
+        return cls(RmaEngine(**_rma_kwargs))

allensdk/brain_observatory/ecephys/ecephys_project_api/warehouse_patches/__init__.py

@@ -0,0 +1,76 @@
+import os
+
+import pandas as pd
+import numpy as np
+
+
+def structure_assignment_fname_map():
+    return {
+        733744647: "732592105_probeB_733744647.csv",
+        733744649: "732592105_probeC_733744649.csv",
+        733744651: "732592105_probeD_733744651.csv",
+        733744653: "732592105_probeE_733744653.csv",
+        733744655: "732592105_probeF_733744655.csv",
+        805579698: "799864342_probeA_805579698.csv",
+        832129149: "829720705_probeA_832129149.csv",
+        837761710: "835479236_probeC_837761710.csv"
+    }
+
+
+def load_structure_assignments(probe_ids=None):
+
+    fnames = structure_assignment_fname_map()
+    dirname = os.path.join(os.path.dirname(__file__), "data")
+
+    output = []
+    for probe_id, fname in fnames.items():
+        if probe_ids is not None and probe_id not in probe_ids:
+            continue
+
+        full_path = os.path.join(dirname, fname)
+        df = pd.read_csv(full_path)
+
+        df["ecephys_probe_id"] = np.zeros([len(df)], dtype=int) + probe_id
+        output.append(df)
+
+    return pd.concat(output)
+
+
+def replace_bad_structure_assignments(
+    channels, 
+    structure_id_key="structure_id", 
+    structure_acronym_key="structure_acronym",
+    inplace=False
+):
+    if not inplace:
+        channels = channels.copy()
+
+    probes = set(channels["ecephys_probe_id"].unique().tolist())
+
+    assignments = load_structure_assignments()
+    reassigned_probes = set(assignments["ecephys_probe_id"].unique().tolist())
+
+    reassigned = channels[channels["ecephys_probe_id"].isin(reassigned_probes)]
+    reassigned = pd.merge(
+        reassigned.reset_index(), assignments,
+        how="left",
+        left_on=["ecephys_probe_id", "local_index"], 
+        right_on=["ecephys_probe_id", "local_index"], 
+        suffixes=["_original", "_corrected"]
+    )
+    reassigned.set_index("id", inplace=True)
+
+    reassigned.drop(columns=[
+        f"{structure_id_key}_original", 
+        f"{structure_acronym_key}_original"
+    ], inplace=True)
+    reassigned.rename(columns={
+        f"{structure_id_key}_corrected": structure_id_key, 
+        f"{structure_acronym_key}_corrected": structure_acronym_key
+    }, inplace=True)
+
+    channels.loc[reassigned.index.values, [structure_id_key, structure_acronym_key]] = \
+        reassigned.loc[:, [structure_id_key, structure_acronym_key]]
+
+    if not inplace:
+        return reassigned