Adding/removing containers from an NWB file

This tutorial explains how to add and remove containers from an existing NWB file and either write the data back to the same file or export the data to a new file.

Adding objects to an NWB file in read/write mode

PyNWB supports adding container objects to an existing NWB file - that is, reading data from an NWB file, adding a container object, such as a new TimeSeries object, and writing the modified NWBFile back to the same file path on disk. To do so:

  1. open the file with an NWBHDF5IO object in read/write mode (mode='r+' or mode='a')
  2. read the NWBFile
  3. add container objects to the NWBFile object
  4. write the modified NWBFile using the same NWBHDF5IO object

For example:

from pynwb import NWBFile, NWBHDF5IO, TimeSeries
import datetime
import numpy as np

# first, write a test NWB file
nwbfile = NWBFile(
    session_description='demonstrate adding to an NWB file',
    identifier='NWB123',
    session_start_time=datetime.datetime.now(datetime.timezone.utc),
)

filename = 'nwbfile.nwb'
with NWBHDF5IO(filename, 'w') as io:
    io.write(nwbfile)

# open the NWB file in r+ mode
with NWBHDF5IO(filename, 'r+') as io:
    read_nwbfile = io.read()

    # create a TimeSeries and add it to the file under the acquisition group
    data = list(range(100, 200, 10))
    timestamps = np.arange(10, dtype=np.float)
    test_ts = TimeSeries(
        name='test_timeseries',
        data=data,
        unit='m',
        timestamps=timestamps
    )
    read_nwbfile.add_acquisition(test_ts)

    # write the modified NWB file
    io.write(read_nwbfile)

# confirm the file contains the new TimeSeries in acquisition
with NWBHDF5IO(filename, 'r') as io:
    read_nwbfile = io.read()
    print(read_nwbfile)

Note

You cannot remove objects from an NWB file using the above method.

Modifying an NWB file in this way has limitations. The destination file path must be the same as the source file path, and it is not possible to remove objects from an NWB file. You can use the NWBHDF5IO.export method, detailed below, to modify an NWB file in these ways.

Warning

NWB datasets that have been written to disk are read as h5py.Dataset objects. Directly modifying the data in these h5py.Dataset objects immediately modifies the data on disk (the NWBHDF5IO.write method does not need to be called and the NWBHDF5IO instance does not need to be closed). Directly modifying datasets in this way can lead to files that do not validate or cannot be opened, so take caution when using this method. Note: only chunked datasets or datasets with maxshape set can be resized. See the h5py chunked storage documentation for more details.

Note

It is not possible to modify the attributes (fields) of an NWB container in memory.

Exporting a written NWB file to a new file path

Use the NWBHDF5IO.export method to read data to an existing NWB file, modify the data, and write the modified data to a new file path. Modifications to the data can be additions or removals of objects, such as TimeSeries objects. This is especially useful if you have raw data and processed data in the same NWB file and you want to create a new NWB file with all of the contents of the original file except for the raw data for sharing with collaborators.

To remove existing containers, use the pop method on any LabelledDict object, such as NWBFile.acquisition, NWBFile.processing, NWBFile.analysis, NWBFile.processing, NWBFile.scratch, NWBFile.devices, NWBFile.stimulus, NWBFile.stimulus_template, NWBFile.electrode_groups, NWBFile.imaging_planes, NWBFile.icephys_electrodes, NWBFile.ogen_sites, NWBFile.lab_meta_data, and ProcessingModule objects.

For example:

# first, create a test NWB file with a TimeSeries in the acquisition group
nwbfile = NWBFile(
    session_description='demonstrate export of an NWB file',
    identifier='NWB123',
    session_start_time=datetime.datetime.now(datetime.timezone.utc),
)
data1 = list(range(100, 200, 10))
timestamps1 = np.arange(10, dtype=np.float)
test_ts1 = TimeSeries(
    name='test_timeseries1',
    data=data1,
    unit='m',
    timestamps=timestamps1
)
nwbfile.add_acquisition(test_ts1)

# then, create a processing module for processed behavioral data
nwbfile.create_processing_module(
    name='behavior',
    description='processed behavioral data'
)
data2 = list(range(100, 200, 10))
timestamps2 = np.arange(10, dtype=np.float)
test_ts2 = TimeSeries(
    name='test_timeseries2',
    data=data2,
    unit='m',
    timestamps=timestamps2
)
nwbfile.processing['behavior'].add(test_ts2)

# write these objects to an NWB file
filename = 'nwbfile.nwb'
with NWBHDF5IO(filename, 'w') as io:
    io.write(nwbfile)

# read the written file
export_filename = 'exported_nwbfile.nwb'
with NWBHDF5IO(filename, mode='r') as read_io:
    read_nwbfile = read_io.read()

    # add a new TimeSeries to the behavior processing module
    data3 = list(range(100, 200, 10))
    timestamps3 = np.arange(10, dtype=np.float)
    test_ts3 = TimeSeries(
        name='test_timeseries3',
        data=data3,
        unit='m',
        timestamps=timestamps3
    )
    read_nwbfile.processing['behavior'].add(test_ts3)

    # use the pop method to remove the original TimeSeries from the acquisition group
    read_nwbfile.acquisition.pop('test_timeseries1')

    # use the pop method to remove a TimeSeries from a processing module
    read_nwbfile.processing['behavior'].data_interfaces.pop('test_timeseries2')

    # call the export method to write the modified NWBFile instance to a new file path
    # the original file is not modified
    with NWBHDF5IO(export_filename, mode='w') as export_io:
        export_io.export(src_io=read_io, nwbfile=read_nwbfile)

# confirm the exported file does not contain TimeSeries with names 'test_timeseries1' or 'test_timeseries2'
# but does contain a new TimeSeries in processing['behavior'] with name 'test_timeseries3'
with NWBHDF5IO(export_filename, 'r') as io:
    read_nwbfile = io.read()
    print(read_nwbfile)
    print(read_nwbfile.processing['behavior'])

Note

TimeIntervals objects, such as NWBFile.epochs, NWBFile.trials, NWBFile.invalid_times, and custom TimeIntervals objects cannot be removed (popped) from NWBFile.intervals.

Warning

Removing an object from an NWBFile may break links and references within the file and across files. This is analogous to having shortcuts/aliases to a file on your filesystem and then deleting the file. Extra caution should be taken when removing heavily referenced items such as Device objects, ElectrodeGroup objects, the electrodes table, and the PlaneSegmentation table.

Exporting with new object IDs

When exporting a read NWB file to a new file path, the object IDs within the original NWB file will be copied to the new file. To make the exported NWB file contain a new set of object IDs, call generate_new_id on your NWBFile object. This will generate a new object ID for the NWBFile object and all of the objects within the NWB file.

export_filename = 'exported_nwbfile.nwb'
with NWBHDF5IO(filename, mode='r') as read_io:
    read_nwbfile = read_io.read()
    read_nwbfile.generate_new_id()

    with NWBHDF5IO(export_filename, mode='w') as export_io:
        export_io.export(src_io=read_io, nwbfile=read_nwbfile)