.. _section-neo-rawio-API: ============= Neo RawIO API ============= .. currentmodule:: neo.rawio For performance and memory consumption reasons, Neo provides a low-level, developer-oriented read-only API for reading different file formats. Neo's full-featured IO modules are built on this, but it is also available for direct use. In brief: - **neo.io** is the user-oriented read/write layer. Reading consists of getting a tree of Neo objects from a data source (file, url, or directory). When reading, all Neo objects are correctly scaled to the correct units. Writing consists of making a set of Neo objects persistent in a file format. - **neo.rawio** is a low-level layer for reading data only. Reading consists of getting NumPy buffers (often int16/int64) of signals/spikes/events. Scaling to real values (microV, times, ...) is done in a second step. Here the underlying objects must be consistent across Blocks and Segments for a given data source. The neo.rawio API is close in spirit to a C API for reading data, but in Python/NumPy. Many, but not all of the file formats supported in :mod:`neo.io` also have a :mod:`neo.rawio` interface. Possible uses of the :mod:`neo.rawio` API are: - fast reading chunks of signals in int16 and do the scaling of units (uV) on a GPU while scaling the zoom. This should improve bandwidth from HD/SSD to RAM and from RAM to GPU memory. - load only a small chunk of data for heavy computations. For instance the spike sorting module tridesclous_ does this. The :mod:`neo.rawio` API is less flexible than :mod:`neo.io` and has some limitations: - read-only - AnalogSignals must have the same characteristics across all Blocks and Segments: ``sampling_rate``, ``shape[1]``, ``dtype`` - AnalogSignals should all have the same value of ``sampling_rate``, otherwise they won't be read at the same time. - Units must have SpikeTrains even if empty across all Block and Segment - Epoch and Event are processed the same way (with ``durations=None`` for Event). For an intuitive comparison of :mod:`neo.io` and :mod:`neo.rawio` see: - :doc:`examples/read_files_neo_io` - :doc:`examples/read_files_neo_rawio` One benefit of the :mod:`neo.rawio` API is that a developer should be able to code a new RawIO class with little knowledge of the Neo tree of objects or of the :mod:`quantities` package. Basic usage =========== .. Download example files .. ipython:: python :suppress: import os.path from urllib.request import urlretrieve url_repo = 'https://web.gin.g-node.org/NeuralEnsemble/ephy_testing_data/raw/master/' for distantfile in ('plexon/File_plexon_2.plx', 'plexon/File_plexon_3.plx', 'blackrock/FileSpec2.3001.nev', 'blackrock/FileSpec2.3001.ns5'): localfile = distantfile.split("/")[1] if not os.path.exists(localfile): urlretrieve(url_repo + distantfile, localfile) First create a reader from a class: .. ipython:: In [1]: from neo.rawio import PlexonRawIO In [2]: reader = PlexonRawIO(filename='File_plexon_3.plx') Then browse the internal header and display information: .. ipython:: In [3]: reader.parse_header() In [4]: reader Out[4]: PlexonRawIO: File_plexon_3.plx nb_block: 1 nb_segment: [1] signal_channels: [V1] spike_channels: [Wspk1u, Wspk2u, Wspk4u, Wspk5u ... Wspk29u Wspk30u Wspk31u Wspk32u] event_channels: [] You get the number of blocks and segments per block. You have information about channels: :attr:`signal_channels`, :attr:`spike_channels`, :attr:`event_channels`. All this information is available in the :attr:`header` dict: .. ipython:: In [5]: for k, v in reader.header.items(): ...: print(k, v) Out[5]: signal_channels [('V1', 0, 1000., 'int16', '', 2.44140625, 0., 0)] event_channels [] nb_segment [1] nb_block 1 spike_channels [('Wspk1u', 'ch1#0', '', 0.00146484, 0., 0, 30000.) ('Wspk2u', 'ch2#0', '', 0.00146484, 0., 0, 30000.) Read chunks of signal data and scale them ----------------------------------------- .. ipython:: In [6]: channel_indexes = None #could be channel_indexes = [0] In [7]: raw_sigs = reader.get_analogsignal_chunk(block_index=0, seg_index=0, ...: i_start=1024, i_stop=2048, ...: channel_indexes=channel_indexes) In [8]: float_sigs = reader.rescale_signal_raw_to_float(raw_sigs, dtype='float64') In [9]: sampling_rate = reader.get_signal_sampling_rate() In [10]: t_start = reader.get_signal_t_start(block_index=0, seg_index=0) In [11]: units = reader.header['signal_channels'][0]['units'] In [12]: raw_sigs.shape, raw_sigs.dtype Out[12]: ((1024, 1), dtype('int16')) In [13]: float_sigs.shape, float_sigs.dtype Out[13]: ((1024, 1), dtype('float64')) In [14]: sampling_rate, t_start, units Out[14]: (1000.0, 0.0, '') There are 3 ways to select a subset of channels: by index (0 based), by id or by name. By index is unambiguous 0 to n-1 (inclusive), whereas for some IOs channel_names (and sometimes channel_ids) are not guaranteed to be unique. In such cases, using names or ids may raise an error. A selected subset of channels which is passed to :func:`get_analog_signal_chunk()`, :func:`get_analog_signal_size()`, or :func:`get_analog_signal_t_start()` has the additional restriction that all such channels must have the same :attr:`t_start` and :attr:`signal_size`. Such subsets of channels may be available in specific RawIOs by using the :func:`get_group_signal_channel_indexes()` method, if the RawIO has defined separate :attr:`group_ids` for each group with those common characteristics. Example with BlackrockRawIO for the recording `FileSpec2.3001`_: .. ipython:: In [15]: from neo.rawio import BlackrockRawIO In [16]: reader = BlackrockRawIO(filename="FileSpec2.3001") In [17]: reader.parse_header() In [18]: raw_sigs = reader.get_analogsignal_chunk(channel_indexes=None) # Take all channels In [19]: raw_sigs1 = reader.get_analogsignal_chunk(channel_indexes=[0, 2, 4]) # Take 0 2 and 4 In [20]: raw_sigs2 = reader.get_analogsignal_chunk(channel_ids=['1', '3', '5']) # Same but with their id (1 based) In [21]: raw_sigs3 = reader.get_analogsignal_chunk(channel_names=['chan1', 'chan3', 'chan5']) # Same but with their name In [22]: raw_sigs1.shape[1], raw_sigs2.shape[1], raw_sigs3.shape[1] Out[22]: (3, 3, 3) Inspect spiking unit channels ----------------------------- Each channel gives a SpikeTrain for each Segment. Note that for many formats a physical channel can have several units after spike sorting. So the number of spike channels could be more than the number of physical channels or signal channels. .. ipython:: In [23]: nb_unit = reader.spike_channels_count() In [24]: print('nb_unit', nb_unit) nb_unit 4 In [25]: for spike_channel_index in range(nb_unit): ....: nb_spike = reader.spike_count(block_index=0, seg_index=0, spike_channel_index=spike_channel_index) ....: print('spike_channel_index', spike_channel_index, 'nb_spike', nb_spike) spike_channel_index 0 nb_spike 259 spike_channel_index 1 nb_spike 234 spike_channel_index 2 nb_spike 218 spike_channel_index 3 nb_spike 253 Get spike timestamps in a defined time range and convert them to spike times ---------------------------------------------------------------------------- .. ipython:: In [26]: spike_timestamps = reader.get_spike_timestamps(block_index=0, seg_index=0, spike_channel_index=0, ....: t_start=0, t_stop=10) In [27]: print(spike_timestamps.shape, spike_timestamps.dtype, spike_timestamps[:5]) (86,) uint32 [ 19312 49298 79301 139290 162170] In [28]: spike_times = reader.rescale_spike_timestamp( spike_timestamps, dtype='float64') In [29]: print(spike_times.shape, spike_times.dtype, spike_times[:5]) (86,) float64 [0.64373333 1.64326667 2.64336667 4.643 5.40566667] Get spike waveforms in a defined time range ------------------------------------------- .. ipython:: In [30]: raw_waveforms = reader.get_spike_raw_waveforms(block_index=0, seg_index=0, spike_channel_index=0, ....: t_start=0, t_stop=10) In [31]: print(raw_waveforms.shape, raw_waveforms.dtype, raw_waveforms[0, 0, :4]) (86, 1, 48) int16 [-209 -224 -74 205] In [32]: float_waveforms = reader.rescale_waveforms_to_float(raw_waveforms, dtype='float32', spike_channel_index=0) In [33]: print(float_waveforms.shape, float_waveforms.dtype, float_waveforms[0,0,:4]) (86, 1, 48) float32 [-52.25 -56. -18.5 51.25] Count events per channel ------------------------ .. ipython:: In [34]: reader = PlexonRawIO(filename='File_plexon_2.plx') In [35]: reader.parse_header() In [36]: nb_event_channel = reader.event_channels_count() In [37]: print('nb_event_channel', nb_event_channel) nb_event_channel 28 In [38]: for chan_index in range(nb_event_channel): ....: nb_event = reader.event_count(block_index=0, seg_index=0, event_channel_index=chan_index) ....: print('chan_index',chan_index, 'nb_event', nb_event) chan_index 0 nb_event 1 chan_index 1 nb_event 0 chan_index 2 nb_event 0 chan_index 3 nb_event 0 ... Read event timestamps and times ------------------------------- .. ipython:: In [39]: ev_timestamps, ev_durations, ev_labels = reader.get_event_timestamps( ....: block_index=0, seg_index=0, event_channel_index=0, ....: t_start=None, t_stop=None) In [40]: print(ev_timestamps, ev_durations, ev_labels) [1268] None ['0'] In [41]: ev_times = reader.rescale_event_timestamp(ev_timestamps, dtype='float64') In [42]: print(ev_times) [ 0.0317] List of implemented formats =========================== See :doc:`rawiolist`. .. _tridesclous: https://github.com/tridesclous/tridesclous .. _FileSpec2.3001: https://gin.g-node.org/NeuralEnsemble/ephy_testing_data/src/master/blackrock