fix: address PR comments

aeon/dj_pipeline/analysis/block_analysis.py

@@ -200,8 +200,8 @@ def make(self, key):
                 use_blob_position = True
 
         # Patch data - TriggerPellet, DepletionState, Encoder (distancetravelled)
-        # For wheel data, downsample to 10Hz
-        final_encoder_hz = 10
+        # For wheel data, downsample to 50Hz
+        final_encoder_hz = 50
         freq = 1 / final_encoder_hz * 1e3  # in ms
 
         maintenance_period = get_maintenance_periods(key["experiment_name"], block_start, block_end)

aeon/dj_pipeline/subject.py

@@ -435,6 +435,10 @@ def make(self, key):
 
 
 def get_pyrat_data(endpoint: str, params: dict = None, **kwargs):
+    """
+    Get data from PyRat API.
+    See docs at: https://swc.pyrat.cloud/api/v3/docs (production)
+    """
     base_url = "https://swc.pyrat.cloud/api/v3/"
     pyrat_system_token = os.getenv("PYRAT_SYSTEM_TOKEN")
     pyrat_user_token = os.getenv("PYRAT_USER_TOKEN")

aeon/dj_pipeline/tracking.py

@@ -167,6 +167,7 @@ def make(self, key):
         stream_reader = getattr(getattr(devices_schema, device_name), "Pose")
 
         # special ingestion case for social0.2 full-pose data (using Pose reader from social03)
+        # fullpose for social0.2 has a different "pattern" for non-fullpose, hence the Pose03 reader
         if key["experiment_name"].startswith("social0.2"):
             from aeon.io import reader as io_reader
             stream_reader = getattr(getattr(devices_schema, device_name), "Pose03")
@@ -198,6 +199,9 @@ def make(self, key):
             # FIXME: the logic below to get "anchor_part" is not robust, it relies on the ordering of the unique parts
             #  but if there are missing frames for the actual anchor part, it will be missed
             #  and another part will be incorrectly chosen as "anchor_part"
+            #  (2024-10-31) - we recently discovered that the parts are not sorted in the same order across frames
+            #               - further exacerbating the flaw in the logic below
+            #  best is to find a robust way to get the anchor part info from the config file for this chunk
             anchor_part = np.unique(identity_position.part)[0]
 
             for part in set(identity_position.part.values):