Version 2.0 is an updated version for Python 3.10 with prior Igor code re-implemented in Python.
Version 1.1 is the source code for:
Eisenman et al. Quantification of bursting and synchrony in cultured hippocampal neurons. J Neurophysiol submitted.
It includes both Python and Igor code as described below. Please cite this article if you use any of the code. License issues are addressed in the License.txt file.
The Python code in this package requires Numpy, Scipy and Pandas to run. Cython is required to compile from source.
The Python code has been tested for the 64 bit versions of Python 2.7 and 3.4 - 3.6 in both Windows (AppVeyor) and Linux (Travis)
The Igor code requires Igor Pro.
If you do not already have Python installed on your system or don't know how to ensure you have the required packages installed, your best bet is to use a complete scientific distrubution such as Anaconda, Enthought Canopy or WinPython.
Wheels for the 64 bit versions of Python 2.7 and Python 3.4 - 3.6 were compiled using Appveyor. These wheels can be installed using a current version of pip:
pip install -U pip # for Python2.7 use: pip install -U https://github.com/lneisenman/burst_sync/releases/download/v1.0/burst_sync-1.0-cp27-none-win_amd64.whl # for Python3.4 use: pip install -U https://github.com/lneisenman/burst_sync/releases/download/v1.0/burst_sync-1.0-cp34-none-win_amd64.whl
Assuming the appropriate compilers and development files for Python are
present, the code can be built from source with python setup.py install.
On Ubuntu you would need the build-essential, python-dev gfortran,
liblapack-dev, libatlas-base-dev and python-pip packages that can
be installed using apt-get.
Note that Igor Pro does not run on Linux.
The code for Robust Gaussian Surprise is based on the R code published with:
Ko et al. Detection of bursts and pauses in spike trains. J Neurosci Methods, 2012:145
Please make sure to cite Ko et al. in addition to Eisenman et al. if you use this code.
The code expects the interspike intervals to be in a Pandas dataframe with the label isi which in turn is in a list that is passed to the code. Multiple dataframes can be passed in the same list. It returns lists of dataframes for the identified bursts and pauses.
import scipy as sp
import pandas as pd
from burst_sync import rgs
isi = sp.stats.expon.rvs(lambda=2, size=1000) # ISI's for Poisson spikes
data = [pd.DataFrame({'isi': isi})]
bursts, pauses = rgs.bp_summary(data, min_spikes=3)The SPIKE synchrony measure is described in:
Kreuz et al. Monitoring spike train synchrony. J Neurophysiol, 2013:1457
Please make sure to cite Kreuz et al. in addition to Eisenman et al. if you use this code.
This is a reimplementation of the code from Kreuz using C and Cython. It assumes that the spike times for each channel are in a numpy array. The parameters ti and tf represent the start and end times of the spike train. The last parameter is the number of samples.
import numpy as np
from burst_sync import spike
ti = 0
tf = 1300
samples = 50
unit1 = np.arange(100, 1201, 100, dtype=float)
unit2 = np.arange(100, 1201, 110, dtype=float)
unit3 = np.arange(100, 1201, 120, dtype=float)
tb, Sb = spike.bivariate_spike_distance(unit1, unit2, ti, tf, samples)
bi_distance = np.average(Sb)
unit_list = [unit1, unit2, unit3]
tm, Sm = spike.multivariate_spike_distance(unit_list, ti, tf, samples)
mv_distance = np.average(Sm)The Spike Time Tiling Coefficent is described in:
Cutts and Eglen. Detecting pairwise correlations in spike trains: an objective comparison of methods and application to the study of retinal waves. J Neurosci, 2014:14288
Please make sure to cite Cutts and Eglen in addition to Eisenman et al. if you use this code.
This implementation uses Cython to access the C code from their manuscript.It assumes that the spike times for each channel are in a numpy array. The parameters dt, ti and tf represent the time step, start time and end time of the spike train. The multivariate_sttc returns a 2-D array whose values represent the bivariate sttc for the corresponding pair of spike trains.
import numpy as np
from burst_sync import sttc
ti = 0
tf = 1300
dt = 1
unit1 = np.arange(100, 1201, 100, dtype=float)
unit2 = np.arange(100, 1201, 110, dtype=float)
unit3 = np.arange(100, 1201, 120, dtype=float)
bi_sttc = sttc.sttc(unit1, unit2, dt, ti, tf)
unit_list = [unit1, unit2, unit3]
mv_sttc = sttc.multivariate_sttc(unit_list, dt, ti, tf)The Global synchrony measure is described in:
Li et al. Synchronization measurement of multiple neuronal populations. J Neurophysiol, 2007:3341 Patel et al. Dynamic changes in neural circuit topology following mild mechanical injury in vitro. Annals of biomedical engineering, 2012:23 Patel et al. Single-neuron NMDA receptor phenotype influences neuronal rewiring and reintegration following traumatic injury. J Neurosci, 2014:4200
Please make sure to cite these authors in addition to Eisenman et al. if you use this code.
This code assumes that the spike times for each channel are in a numpy array. The parameter tf represent the end time of the data. of the spike train.
import numpy as np
from burst_sync import global_sync as gs
tf = 1300
unit1 = np.arange(100, 1201, 100, dtype=float)
unit2 = np.arange(100, 1201, 110, dtype=float)
unit3 = np.arange(100, 1201, 120, dtype=float)
unit_list = [unit1, unit2, unit3]
sync = gs.calc_global_sync(unit_list, tf)The Tcrit method is described in:
Wagenaar et al. An extremely rich repertoire of bursting patterns during the development of cortical cultures. BMC Neurosci, 2006:11
The ISI_N method is described in:
Bakkum et al. Parameters for burst detection. Front Comput Neurosci, 2013:193
The B Statistic is described in:
Tiesinga and Sejnowski. Rapid temporal modulation of synchrony by competition in cortical interneuron networks. Neural Comput, 2004:251 Bogaard et al. Interaction of cellular and network mechanisms in spatiotemporal pattern formation in neuronal networks. J Neurosci, 2009:1677
Please make sure to cite the relevant authors in addition to Eisenman et al. if you use any of this code.
This code was written for Igor Pro. The Igor code and a demo experiment are included in the corresponding folder. Instructions are included in the demo experiment.
The boilerplate code for this project was created using PyScaffold.
AppVeyor and Travis configurations were based on the demo projects created by Oliver Grisel and Rob McGibbon