create intervalset from full dataframe with metadata when loading in an nwb file to ensure it is aligned correctly. also try an initial sort of an input dataframe to preserve metadata

Author: sjvenditto

pynapple/core/interval_set.py

@@ -197,6 +197,12 @@ def __init__(
             ), """
                 DataFrame must contain columns name "start" and "end" for start and end times.                   
                 """
+            # try sorting the DataFrame by start times, preserving its end pair, as an effort to preserve metadata
+            # since metadata would be dropped if starts and ends are sorted separately
+            # note that if end times are still not sorted, metadata will be dropped
+            if np.any(start["start"].diff() < 0):
+                start = start.sort_values("start").reset_index(drop=True)
+
             metadata = start.drop(columns=["start", "end"])
             end = start["end"].values.astype(np.float64)
             start = start["start"].values.astype(np.float64)

pynapple/io/interface_nwb.py

@@ -85,10 +85,9 @@ def _make_interval_set(obj, **kwargs):
         df = obj.to_dataframe()
 
         if hasattr(df, "start_time") and hasattr(df, "stop_time"):
-            data = nap.IntervalSet(start=df["start_time"], end=df["stop_time"])
-            if df.shape[1] > 2:
-                metadata = df.drop(columns=["start_time", "stop_time"])
-                data.set_info(metadata)
+            df = df.rename(columns={"start_time": "start", "stop_time": "end"})
+            # create from full dataframe to ensure that metadata is associated correctly
+            data = nap.IntervalSet(df)
             return data
 
     else: