py-mea-axion 0.3.0

End-to-end MEA data analysis for Axion Biosystems instruments

py-mea-axion

End-to-end analysis of multi-electrode array (MEA) recordings from Axion Biosystems instruments, in Python.

py-mea-axion reads raw .spk binary files and carries the analysis through spike metrics, burst detection, network burst detection, and synchrony measurement — all the way to statistical comparisons and publication-ready figures.

---

Installation

Recommended: new conda environment

conda create -n mea python=3.11
conda activate mea
pip install py-mea-axion

Or into an existing environment

pip install py-mea-axion

For development (editable install from a local clone)

git clone https://github.com/Molecularbiologyworld/py-mea-axion.git
cd py-mea-axion
pip install -e .

Requirements: Python ≥ 3.10, numpy, scipy, pandas, matplotlib, pingouin, statsmodels.

---

Quickstart

Command-line interface

The fastest way to get results without writing any Python:

# Quick per-well summary printed to the terminal
mea-axion summary recording.spk --fs-override 12500

# Full pipeline — writes CSVs + figures to results/
mea-axion run recording.spk --out results/ --fs-override 12500

# With a plate map (adds condition labels to outputs)
mea-axion run recording.spk --metadata plate_map.csv --out results/ --fs-override 12500

# Only CSVs, skip figure generation
mea-axion run recording.spk --out results/ --no-figures --fs-override 12500

# Restrict to specific wells
mea-axion summary recording.spk --wells A1 B1 C1 --fs-override 12500

# Analyse only a time window (300-600 s)
mea-axion run recording.spk --time-start 300 --time-end 600 --out results/ --fs-override 12500

Run mea-axion --help or mea-axion run --help for the full option list.

---

Python API

from py_mea_axion import MEAExperiment

exp = MEAExperiment(
    "recording.spk",
    metadata="plate_map.csv",   # optional: well_id, condition, DIV, replicate_id
    fs_override=12500,
    time_start_s=300.0,         # optional: crop to a time window before analysis
    time_end_s=600.0,
    active_threshold_hz=5/60,   # 5 spikes/min, matching NeuralMetric Tools default
).run()

# Tabular results
exp.spike_metrics      # per-electrode: MFR, ISI stats, active flag
exp.burst_table        # per-burst: start/end time, duration, spike count
exp.well_summary       # per-well: n_active, mean MFR, burst rate, STTC
exp.excluded_wells     # wells dropped for having too few active electrodes

# One-liner export
exp.to_csv("results.csv")        # well_summary merged with metadata

# Statistics
res  = exp.compare("mean_mfr_active_hz")        # Mann-Whitney / Kruskal-Wallis
coef = exp.longitudinal("mean_mfr_active_hz")   # mixed-effects model

# Figures
exp.plot_heatmap("A1")                       # electrode MFR heatmap
exp.plot_raster("A1")                        # spike raster + burst overlays
exp.plot_sttc("A1")                          # pairwise STTC matrix
exp.plot_network_timeline("A1")              # network burst Gantt chart
exp.plot_trajectory("mean_mfr_active_hz")    # longitudinal mean ± SEM

Plate-map CSV format

well_id,condition,DIV,replicate_id
A1,WT,14,rep1
A2,KD,14,rep1
B1,WT,14,rep2
B2,KD,14,rep2

---

Key parameters

Burst detection (burst_kwargs)

Parameter	Default	Description
algorithm	isi_threshold	Detection algorithm
max_isi_s	0.1 s	Maximum within-burst ISI
min_spikes	5	Minimum spikes per burst
min_ibi_s	0.0 s	Minimum inter-burst interval (0 = no merging)

Network burst detection (network_kwargs)

Parameter	Default	Description
algorithm	combined_isi	Merges all electrode spike trains, matches NeuralMetric Tools
max_isi_s	0.1 s	Maximum within-network-burst ISI
min_spikes	50	Minimum spikes in combined train
participation_threshold	0.35	Minimum fraction of electrodes that must fire

Pass these via burst_kwargs / network_kwargs in MEAExperiment.

Analysis window

Use time_start_s / time_end_s in the Python API or --time-start / --time-end in the CLI to analyse only a sub-window of the recording. Spikes outside the window are excluded before metrics, burst detection, network detection, and STTC are computed. Internally, the retained timestamps are shifted so that the analysis window starts at t=0.

---

Note on Axion .spk files

Some recordings lack a BlockVectorHeader (common with certain firmware versions). In this case the sampling frequency is inferred automatically; a warning is printed. If the inferred frequency is wrong, override it explicitly:

exp = MEAExperiment("recording.spk", fs_override=12500)

mea-axion run recording.spk --fs-override 12500

---

Running the tests

pip install -e ".[dev]"
pytest

397 tests covering all modules.

---

Citing

If you use py-mea-axion in your research, please cite:

[manuscript in preparation]

---

License

MIT — see LICENSE.