Merge pull request #8 from catalystneuro/epoch

Epoch

src/jadhav_lab_to_nwb/olson_2024/__init__.py

@@ -1,5 +1,8 @@
 from .olson_2024_behavior_interface import Olson2024BehaviorInterface
+from .olson_2024_video_interface import Olson2024VideoInterface
+from .olson_2024_dlc_interface import Olson2024DeepLabCutInterface
 from .olson_2024_spike_gadgets_recording_interface import Olson2024SpikeGadgetsRecordingInterface
 from .olson_2024_sorting_interface import Olson2024SortingInterface
 from .olson_2024_spike_gadgets_lfp_interface import Olson2024SpikeGadgetsLFPInterface
-from .olson_2024_nwbconverter import Olson2024NWBConverter
+from .olson_2024_epoch_interface import Olson2024EpochInterface
+from .olson_2024_nwbconverter import Olson2024NWBConverter, get_start_datetime

src/jadhav_lab_to_nwb/olson_2024/olson_2024_convert_session.py

@@ -1,6 +1,6 @@
 """Primary script to run to convert an entire session for of data using the NWBConverter."""
 from pathlib import Path
-import datetime
+from datetime import datetime
 from zoneinfo import ZoneInfo
 import shutil
 
@@ -9,37 +9,36 @@
 )  # TODO: remove after this issue gets fixed: https://github.com/catalystneuro/neuroconv/issues/1143
 from neuroconv.utils import load_dict_from_file, dict_deep_update
 
-from jadhav_lab_to_nwb.olson_2024 import Olson2024NWBConverter
+from jadhav_lab_to_nwb.olson_2024 import Olson2024NWBConverter, get_start_datetime
 
 
-def session_to_nwb(data_dir_path: str | Path, output_dir_path: str | Path, stub_test: bool = False):
+def session_to_nwb(
+    data_dir_path: str | Path, subject_id: str, session_id: str, output_dir_path: str | Path, stub_test: bool = False
+):
 
     data_dir_path = Path(data_dir_path)
     output_dir_path = Path(output_dir_path)
     if stub_test:
         output_dir_path = output_dir_path / "nwb_stub"
     output_dir_path.mkdir(parents=True, exist_ok=True)
+    session_folder_path = data_dir_path / f"{subject_id}_{session_id}"
+    nwbfile_path = output_dir_path / f"sub-{subject_id}_ses-{session_id}.nwb"
 
-    session_id = "sample_session"
-    subject_id = "SL18"
-    nwbfile_path = output_dir_path / f"{session_id}.nwb"
+    # Get epoch info
+    epoch_folder_paths = list(session_folder_path.glob(rf"{session_folder_path.name}_S[0-9][0-9]_F[0-9][0-9]_*"))
+    epoch_folder_paths = sorted(epoch_folder_paths)
 
     source_data = dict()
     conversion_options = dict()
 
     # Add Ephys
-    file_path = (
-        data_dir_path
-        / "SL18_D19"
-        / "SL18_D19_S01_F01_BOX_SLP_20230503_112642"
-        / "SL18_D19_S01_F01_BOX_SLP_20230503_112642.rec"
-    )
-    source_data.update(dict(Recording=dict(file_path=file_path)))
+    file_paths = [epoch_folder_path / f"{epoch_folder_path.name}.rec" for epoch_folder_path in epoch_folder_paths]
+    source_data.update(dict(Recording=dict(file_paths=file_paths)))
     conversion_options.update(dict(Recording=dict(stub_test=stub_test)))
 
     # Add Sorting
-    spike_times_folder_path = data_dir_path / "SL18_D19" / "SL18_D19.SpikesFinal"
-    unit_stats_folder_path = data_dir_path / "SL18_D19" / "SL18_D19.ExportedUnitStats"
+    spike_times_folder_path = session_folder_path / f"{session_folder_path.name}.SpikesFinal"
+    unit_stats_folder_path = session_folder_path / f"{session_folder_path.name}.ExportedUnitStats"
     source_data.update(
         dict(
             Sorting=dict(spike_times_folder_path=spike_times_folder_path, unit_stats_folder_path=unit_stats_folder_path)
@@ -48,35 +47,41 @@ def session_to_nwb(data_dir_path: str | Path, output_dir_path: str | Path, stub_
     conversion_options.update(dict(Sorting=dict()))
 
     # Add LFP
-    folder_path = data_dir_path / "SL18_D19" / "SL18_D19.LFP"
+    folder_path = session_folder_path / f"{session_folder_path.name}.LFP"
     source_data.update(dict(LFP=dict(folder_path=folder_path)))
     conversion_options.update(dict(LFP=dict(stub_test=stub_test)))
 
+    # Add DLC
+    dlc_folder_path = session_folder_path / f"{session_folder_path.name}.DLC"
+    file_paths = [file_path for file_path in dlc_folder_path.glob(r"*.csv") if not (file_path.name.startswith("._"))]
+    source_data.update(dict(DeepLabCut=dict(file_paths=file_paths, subject_name=subject_id)))
+    conversion_options.update(dict(DeepLabCut=dict()))
+
     # Add Video
-    file_paths = [
-        data_dir_path
-        / "SL18_D19"
-        / "SL18_D19_S01_F01_BOX_SLP_20230503_112642"
-        / "SL18_D19_S01_F01_BOX_SLP_20230503_112642.1.h264"
-    ]
+    file_paths = []
+    for epoch_folder_path in epoch_folder_paths:
+        file_path = epoch_folder_path / f"{epoch_folder_path.name}.1.h264"
+        file_paths.append(file_path)
     source_data.update(dict(Video=dict(file_paths=file_paths)))
     conversion_options.update(dict(Video=dict()))
 
-    # Add DLC
-    file_path = "/Volumes/T7/CatalystNeuro/Jadhav/SubLearnProject/SL18_D19/SL18_D19.DLC/SL18_D19_S01_F01_BOX_SLP_20230503_112642.1DLC_resnet50_SubLearnSleepBoxRedLightJun26shuffle1_100000.csv"
-    source_data.update(dict(DeepLabCut=dict(file_path=file_path, subject_name=subject_id)))
-    conversion_options.update(dict(DeepLabCut=dict()))
-
     # Add Behavior
-    folder_path = "/Volumes/T7/CatalystNeuro/Jadhav/SubLearnProject/SL18_D19/SL18_D19.DIO"
+    folder_path = session_folder_path / f"{session_folder_path.name}.DIO"
     source_data.update(dict(Behavior=dict(folder_path=folder_path)))
     conversion_options.update(dict(Behavior=dict()))
 
+    # Add Epoch
+    source_data.update(dict(Epoch=dict(epoch_folder_paths=epoch_folder_paths)))
+    conversion_options.update(dict(Epoch=dict()))
+
     converter = Olson2024NWBConverter(source_data=source_data)
+    metadata = converter.get_metadata()
 
     # Add datetime to conversion
-    metadata = converter.get_metadata()
-    metadata["NWBFile"]["session_start_time"] = datetime.datetime(2023, 5, 3, 11, 26, 42, tzinfo=ZoneInfo("US/Eastern"))
+    session_start_time = get_start_datetime(epoch_folder_paths[0].name)
+    est = ZoneInfo("US/Eastern")
+    session_start_time = session_start_time.replace(tzinfo=est)
+    metadata["NWBFile"]["session_start_time"] = session_start_time
 
     # Update default metadata with the editable in the corresponding yaml file
     editable_metadata_path = Path(__file__).parent / "olson_2024_metadata.yaml"
@@ -97,8 +102,13 @@ def session_to_nwb(data_dir_path: str | Path, output_dir_path: str | Path, stub_
     if output_dir_path.exists():
         shutil.rmtree(output_dir_path, ignore_errors=True)
 
+    # Example Session
+    subject_id = "SL18"
+    session_id = "D19"
     session_to_nwb(
         data_dir_path=data_dir_path,
+        subject_id=subject_id,
+        session_id=session_id,
         output_dir_path=output_dir_path,
         stub_test=stub_test,
     )

src/jadhav_lab_to_nwb/olson_2024/olson_2024_dlc_interface.py

@@ -0,0 +1,61 @@
+"""Primary class for converting experiment-specific behavioral video."""
+from pynwb.file import NWBFile
+from pydantic import FilePath
+from typing import Optional
+
+from neuroconv.utils import DeepDict, dict_deep_update
+from neuroconv.basedatainterface import BaseDataInterface
+from neuroconv.datainterfaces import DeepLabCutInterface
+
+from .utils.utils import get_epoch_name
+
+
+class Olson2024DeepLabCutInterface(BaseDataInterface):
+    """DeepLabCut interface for olson_2024 conversion"""
+
+    keywords = ("DLC",)
+
+    def __init__(
+        self,
+        file_paths: list[FilePath],
+        config_file_paths: Optional[list[FilePath]] = None,
+        subject_name: str = "ind1",
+        verbose: bool = True,
+    ):
+        # file_paths must be sorted in the order that the videos were recorded
+        assert len(file_paths) > 0, "At least one file path must be provided."
+        if config_file_paths is None:
+            config_file_paths = [None] * len(file_paths)
+        assert len(file_paths) == len(
+            config_file_paths
+        ), "The number of file paths must match the number of config file paths."
+        dlc_interfaces = []
+        for file_path, config_file_path in zip(file_paths, config_file_paths):
+            dlc_interface = DeepLabCutInterface(
+                file_path=file_path,
+                config_file_path=config_file_path,
+                subject_name=subject_name,
+                verbose=verbose,
+            )
+            dlc_interfaces.append(dlc_interface)
+        self.dlc_interfaces = dlc_interfaces
+
+    def get_metadata(self) -> DeepDict:
+        metadata = super().get_metadata()
+        for dlc_interface in self.dlc_interfaces:
+            metadata = dict_deep_update(metadata, dlc_interface.get_metadata())
+        return metadata
+
+    def get_metadata_schema(self) -> DeepDict:
+        metadata_schema = super().get_metadata_schema()
+        for dlc_interface in self.dlc_interfaces:
+            metadata_schema = dict_deep_update(metadata_schema, dlc_interface.get_metadata_schema())
+        return metadata_schema
+
+    def add_to_nwbfile(self, nwbfile: NWBFile, metadata: dict):
+        for dlc_interface in self.dlc_interfaces:
+            file_path = dlc_interface.source_data["file_path"]
+            epoch_name = get_epoch_name(name=file_path.name)
+            dlc_interface.add_to_nwbfile(
+                nwbfile=nwbfile, metadata=metadata, container_name=f"PoseEstimation_{epoch_name}"
+            )

src/jadhav_lab_to_nwb/olson_2024/olson_2024_epoch_interface.py

@@ -0,0 +1,93 @@
+"""Primary class for converting experiment-specific behavior."""
+from pynwb.file import NWBFile
+from pydantic import DirectoryPath
+import numpy as np
+
+from neuroconv.basedatainterface import BaseDataInterface
+from .tools.spikegadgets import readCameraModuleTimeStamps
+from .utils.utils import get_epoch_name
+
+
+class Olson2024EpochInterface(BaseDataInterface):
+    """Epoch interface for olson_2024 conversion"""
+
+    keywords = ("behavior",)
+
+    def __init__(self, epoch_folder_paths: list[DirectoryPath]):
+        super().__init__(epoch_folder_paths=epoch_folder_paths)
+
+    def get_metadata_schema(self):
+        metadata_schema = super().get_metadata_schema()
+        metadata_schema["properties"]["Epochs"] = {
+            "description": "Metadata for each epoch",
+            "type": "array",
+            "items": {
+                "type": "object",
+                "properties": {
+                    "name": {"type": "string"},
+                    "task_name": {"type": "string"},
+                    "led_configuration": {"type": "string"},
+                    "led_list": {"type": "array", "items": {"type": "string"}},
+                    "led_positions": {"type": "array", "items": {"type": "string"}},
+                },
+                "required": ["name", "task_name", "led_configuration", "led_list", "led_positions"],
+            },
+        }
+        metadata_schema["properties"]["Tasks"] = {
+            "description": "Metadata for each task",
+            "type": "array",
+            "items": {
+                "type": "object",
+                "properties": {
+                    "name": {"type": "string"},
+                    "description": {"type": "string"},
+                    "camera_id": {"type": "string"},
+                },
+                "required": ["name", "description", "camera_id"],
+            },
+        }
+        return metadata_schema
+
+    def add_to_nwbfile(self, nwbfile: NWBFile, metadata: dict):
+        epoch_folder_paths = self.source_data["epoch_folder_paths"]
+        nwbfile.add_epoch_column(name="epoch_name", description="Full name of the epoch")
+        nwbfile.add_epoch_column(name="epoch_id", description="Epoch ID")
+        nwbfile.add_epoch_column(name="frag_id", description="Frag ID")  # TODO: What is a frag ID?
+        nwbfile.add_epoch_column(name="env_id", description="Environment ID")
+        nwbfile.add_epoch_column(name="task_id", description="Task ID")
+        nwbfile.add_epoch_column(name="task_name", description="Full name of the task")
+        nwbfile.add_epoch_column(name="task_description", description="Description of the task")
+        nwbfile.add_epoch_column(name="camera_id", description="Camera ID")
+        nwbfile.add_epoch_column(name="led_configuration", description="LED configuration")
+        nwbfile.add_epoch_column(name="led_list", description="Comma-separated list of LEDs")
+        nwbfile.add_epoch_column(name="led_positions", description="Comma-separated list of LED positions")
+        for epoch_folder_path in epoch_folder_paths:
+            epoch_name = get_epoch_name(epoch_folder_path.name)
+            epoch_id, frag_id, env_id, task_id = epoch_name.split("_")
+            epoch_metadata = next(meta for meta in metadata["Epochs"] if meta["name"] == epoch_id)
+            task_name = epoch_metadata["task_name"]
+            task_metadata = next(meta for meta in metadata["Tasks"] if meta["name"] == task_name)
+            task_description = task_metadata["description"]
+            camera_id = task_metadata["camera_id"]
+            led_configuration = epoch_metadata["led_configuration"]
+            led_list = ",".join(epoch_metadata["led_list"])
+            led_positions = ",".join(epoch_metadata["led_positions"])
+            video_timestamps_file_path = epoch_folder_path / f"{epoch_folder_path.name}.1.videoTimeStamps"
+            timestamps, _ = readCameraModuleTimeStamps(video_timestamps_file_path)
+            start_time = timestamps[0]
+            stop_time = timestamps[-1]
+            nwbfile.add_epoch(
+                start_time=start_time,
+                stop_time=stop_time,
+                epoch_name=epoch_name,
+                epoch_id=epoch_id,
+                frag_id=frag_id,
+                env_id=env_id,
+                task_id=task_id,
+                task_name=task_name,
+                task_description=task_description,
+                camera_id=camera_id,
+                led_configuration=led_configuration,
+                led_list=led_list,
+                led_positions=led_positions,
+            )

src/jadhav_lab_to_nwb/olson_2024/olson_2024_metadata.yaml

@@ -37,9 +37,7 @@ Ecephys:
       location: Right Subiculum
       device: ProbeNameTBD
       nTrodes: [44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59]
-  ElectricalSeries:
-    - name: ElectricalSeries
-      description: Raw acquisition of extracellular electrophysiology data recorded by SpikeGadgets.
+  ElectricalSeries_description: Raw acquisition of extracellular electrophysiology data recorded by SpikeGadgets.
   LFP:
     ElectricalSeries:
       - name: ElectricalSeriesLFP
@@ -49,9 +47,7 @@ Behavior:
   Module:
     name: behavior
     description: Behavioral data recorded during foraging shuttle task, in which the rat must shuttle between home and distal reward locations (reward well).
-  Videos:
-    - name: Video SL18_D19_S01_F01_BOX_SLP_20230503_112642.1
-      description: Video of the rat in the box.
+  Video_description: Video of the rat in the box.
   Events:
   - id: ECU_Din1
     name: reward_well_1
@@ -125,3 +121,68 @@ Behavior:
   - id: ECU_Dout24
     name: barrier_8
     description: Whenever Barrier 8 is placed in the box
+
+Epochs:
+  - name: S01
+    task_name: Sleep
+    led_configuration: single
+    led_list:
+      - redled
+    led_positions:
+      - center
+  - name: S02
+    task_name: HomeAltVisitAll
+    led_configuration: left/right
+    led_list:
+      - redled
+      - greenled
+    led_positions:
+      - right
+      - left
+  - name: S03
+    task_name: Sleep
+    led_configuration: single
+    led_list:
+      - redled
+    led_positions:
+      - center
+  - name: S04
+    task_name: HomeAltVisitAll
+    led_configuration: left/right
+    led_list:
+      - redled
+      - greenled
+    led_positions:
+      - right
+      - left
+  - name: S05
+    task_name: Sleep
+    led_configuration: single
+    led_list:
+      - redled
+    led_positions:
+      - center
+  - name: S06
+    task_name: HomeAltVisitAll
+    led_configuration: left/right
+    led_list:
+      - redled
+      - greenled
+    led_positions:
+      - right
+      - left
+  - name: S07
+    task_name: Sleep
+    led_configuration: single
+    led_list:
+      - redled
+    led_positions:
+      - center
+
+Tasks:
+  - name: Sleep
+    description: The animal sleeps in a small empty box.
+    camera_id: "0"
+  - name: HomeAltVisitAll
+    description: Shuttle task between home and 4 destinations.
+    camera_id: "1"