ap-features 2026.2.0

Package to compute features of traces from action potential models

Action Potential features

ap_features is package for computing features of action potential traces. This includes chopping, background correction and feature calculations.

Parts of this library is written in numba and is therefore highly performant. This is useful if you want to do feature calculations on a large number of traces.

Quick start

import matplotlib.pyplot as plt
import numpy as np
from scipy.integrate import solve_ivp

import ap_features as apf

time = np.linspace(0, 999, 1000)
res = solve_ivp(
    apf.testing.fitzhugh_nagumo,
    [0, 1000],
    [0.0, 0.0],
    t_eval=time,
)
trace = apf.Beats(y=res.y[0, :], t=time)
print(f"Number of beats: {trace.num_beats}")
print(f"Beat rates: {trace.beat_rates}")

# Get a list of beats
beats = trace.beats
# Pick out the second beat
beat = beats[1]

# Compute features
print(f"APD30: {beat.apd(30):.3f}s, APD80: {beat.apd(80):.3f}s")
print(f"cAPD30: {beat.capd(30):.3f}s, cAPD80: {beat.capd(80):.3f}s")
print(f"Time to peak: {beat.ttp():.3f}s")
print(f"Decay time from max to 90%: {beat.tau(a=0.1):.3f}s")

Number of beats: 5
Beat rates: [779.2207792207793, 769.2307692307693, 779.2207792207793, 759.493670886076]
APD30: 37.823s, APD80: 56.564s
cAPD30: 88.525s, cAPD80: 132.387s
Time to peak: 21.000s
Decay time from max to 90%: 53.618s

Install

Install the package with pip

python -m pip install ap_features

See installation instructions for more options.

Available features

The list of currently implemented features are as follows

• Action potential duration (APD)
• Corrected action potential duration (cAPD)
• Decay time (Time for the signal amplitude to go from maximum to (1 - a) * 100 % of maximum)
• Time to peak (ttp)
• Upstroke time (time from (1-a)*100 % signal amplitude to peak)
• Beating frequency
• APD up (The duration between first intersections of two APD lines)
• Maximum relative upstroke velocity
• Maximum upstroke velocity
• APD integral (integral of the signals above the APD line)

Documentation

Documentation is hosted at GitHub pages: https://computationalphysiology.github.io/ap_features/

Note that the documentation is written using jupyterbook and contains an interactive demo

License

• Free software: LGPLv2.1

Source Code

• https://github.com/ComputationalPhysiology/ap_features