AP_begin_indices seems to be disturbed by spontaneous firing

[SzaBoglarka]

Hi,
We are having problems with the feature extraction of the trace below.
It seems to me that AP_begin_indices are not detected before the start of the stimulus, but AP_end_indices are.
This issue results in an unequal number of AP_begin_indices (red) and AP_end_indices (dark blue) and causes several problems with other features.
The most concerning results are the negative feature values for AP_duration and AP_rise_time.

Data:
time.csv
voltage.csv

Script:

import numpy
import pandas as pd 
import matplotlib.pyplot as plt
import efel


# print the efel version
print("\n","Efel version:",efel.__version__, "\n")

# load the time and voltage data
time_csv = pd.read_csv("time.csv", header=None, index_col=False)
time_data = time_csv[0].tolist()
voltage_csv = pd.read_csv("voltage.csv", header=None, index_col=False)
voltage_data = voltage_csv[0].tolist()

# create the trace
trace = {}
trace["T"] = time_data
trace["V"] = voltage_data
trace["stim_start"] = [200]
trace["stim_end"] = [200 + 200]
traces = [trace]

# extract the features from the trace
traces_results = efel.getFeatureValues(traces, ["AP_begin_indices", "AP_end_indices", "AP_duration", "AP_rise_time"])

# print the extracted features
print("AP_begin_indices\n:", traces_results[0]['AP_begin_indices'])
print("number of AP_begin_indices:",len(traces_results[0]['AP_begin_indices']), "\n")
print("AP_end_indices\n", traces_results[0]['AP_end_indices'])
print("number of AP_end_indices:",len(traces_results[0]['AP_end_indices']), "\n")
print("AP_durations:\n", traces_results[0]['AP_duration'])
print("number of AP_durations:",len(traces_results[0]['AP_duration']), "\n")
print("AP_rise_time\n", traces_results[0]['AP_rise_time'])
print("number of AP_rise_time:",len(traces_results[0]['AP_rise_time']), "\n")

# plot the data
plt.figure()
plt.plot(time_data,voltage_data)

# mark the AP begin (red) and end (dark blue) indices on the figure
h_dt = 0.025
x = traces_results[0]['AP_begin_indices']/10
votage_indices = [int(i/h_dt) for i in x]       # rescaling with h.dt=0.025
y = [voltage_data[j] for j in votage_indices]
plt.plot(x, y, 'o', color='red', markersize=4)

x = traces_results[0]['AP_end_indices']/10
votage_indices = [int(i/h_dt) for i in x]
y = [voltage_data[j] for j in votage_indices]
plt.plot(x, y, 'o', color='darkblue', markersize=4)

plt.show()

[anilbey]

Thanks @SzaBoglarka for reporting this important issue. We will address it and inform you.

[AurelienJaquier]

Hi @SzaBoglarka ! I am currently fixing this Issue. Is that alright if I use your trace data to add a test in eFel in order to avoid this kind of issue in the future?

[SzaBoglarka]

Hi @AurelienJaquier! Of course, it's alright.
By the way, the trace was generated with a public cell model available here: https://modeldb.science/114047.

[AurelienJaquier]

Thank you very much!
I solved this issue with PR #353 . You can update your efel version. Do not hesitate to contact us if the issue persists or if you encounter another one