ezmsg-neo#

Load and stream data from neo files into ezmsg.

Installation#

pip install ezmsg-neo

Or install the latest development version:

pip install git+https://github.com/ezmsg-org/ezmsg-neo@dev

Dependencies#

Usage#

ezmsg-neo modules are available under import ezmsg.neo

Add the NeoIteratorUnit to your ezmsg graph as a data source. You may be interested in other ezmsg extensions for processing the data, such as ezmsg-sigproc and ezmsg-event.

import ezmsg.core as ez
from ezmsg.neo.source import NeoIteratorUnit
from ezmsg.util.messages.key import FilterOnKey
from ezmsg.util.debuglog import DebugLog
from ezmsg.event.rate import EventRate


comps = {
    "NEO": NeoIteratorUnit(filepath="path/to/file", chunk_dur=0.05),
    "FILTER": FilterOnKey(key="spike"),
    "RATE": EventRate(bin_duration=0.05),
    "LOG": DebugLog()  # Print the output to the console
}
conns = (
    (comps["NEO"].OUTPUT_SIGNAL, comps["FILTER"].INPUT_SIGNAL),
    (comps["FILTER"].OUTPUT_SIGNAL, comps["RATE"].INPUT_SIGNAL),
    (comps["RATE"].OUTPUT_SIGNAL, comps["LOG"].INPUT),
)
ez.run(components=comps, connections=conns)

Standalone#

The NeoIterator class can be used outside ezmsg for offline processing:

from ezmsg.neo.source import NeoIterator, NeoIteratorSettings
from ezmsg.util.messages.axisarray import AxisArray

settings = NeoIteratorSettings(filepath="data.nev", chunk_dur=0.05)
neo_iter = NeoIterator(settings)

# Filter messages by key (e.g., "ns3" for analog signals, "spike" for spikes)
sig_msgs = [msg for msg in neo_iter if msg.key.startswith("ns")]

# Concatenate all chunks into a single AxisArray
full_signal = AxisArray.concatenate(*sig_msgs, dim="time")

NeoIterator Messages#

The NeuoIteratorUnit and NeoIterator objects, when called, return AxisArray messages with a key attribute identifying the signal type:

  • Analog signals - Key matches the stream name (e.g., "ns3"). Data shape: (n_samples, n_channels)

  • Spike trains - Key is "spike". Data is a sparse array with shape (n_units, n_samples)

  • Events - Key is "events". Data contains event labels and its time axis is irregular (CoordinateAxis) with timestamps associated with each event.

Development#

  1. Clone this repo and cd into it

  2. uv sync to setup your environment

  3. uv run pytest tests to run the tests

  4. (Optional) Install pre-commit hooks: uv run pre-commit install

Documentation#

Contents:

Indices and tables#