spike_finder.py

import openephys as oe
import numpy as np
import matplotlib.pyplot as plt
import os
import psutil
from scipy import signal
import numpy as np


def bandpass_data(data, fs= 30000, highcut=6000, lowcut = 300, order = 3):
	nyq = 0.5*fs
	low = lowcut/nyq 
	high = highcut/nyq 
	sos = signal.butter(3, [low, high], analog=False, btype='band', output = 'sos')
	y = signal.sosfiltfilt(sos, data)
	return y

def single_chan_spike_find(bp_data, *, window=0.003, fs=30000):
	w_fs = fs*window
	all_spike_times = []
	all_spike_vals = []
	spike_times = []
	sd = np.median(abs(bp_data)/0.6745)
	data_times = []
	prev_spike = -30
	for index, j in enumerate(bp_data):
		if j < -4*sd and index-prev_spike > 30:
			data_times.append(index)
			prev_spike = index
	spike_times.append(data_times)
	return spike_times

def find_spikes(dataloc, dtype, noc):
	if dtype == 'dat':
		load_dat(dataloc, noc)
	elif dtype == 'continuous':
		load_continuous(dataloc)

available_cpu_count = len(psutil.Process().cpu_affinity())
os.environ["MKL_NUM_THREADS"] = str(available_cpu_count)



chans = [1, 17, 8, 24, 16, 2, 29, 32, 7, 26, 3, 15, 21, 19, 11, 23, 14, 12, 28, 30, 6, 18, 9, 13, 22, 25, 5, 27, 10, 4, 31, 20]

home_dir = '/home/camp/warnert/working/Recordings/190211/2019-02-11_16-35-46'

all_spike_times = []
for chan in chans:
	channel_base = os.path.join(home_dir, '100_CH%d.continuous' % chan)

	chan_rec = oe.loadContinuous2(channel_base)

	data =chan_rec['data']
	bp_data = bandpass_data(data)
	spike_times = spike_find(bp_data)
	all_spike_times.append(spike_times)

all_spike_times = np.array(all_spike_times)
np.save(os.path.join(home_dir, 'channel_spike_times.npy'), all_spike_times)