windolene_example_rasters_andpsth_savetonpyformatrix.py

import pickle
from pathlib import Path
from scipy.ndimage import gaussian_filter1d  # Import Gaussian filter for smoothing

import pandas as pd
import tensorflow as tf
import neo
import numpy as np
# from sklearn.metrics import confusion_matrix
# import matplotlib.pyplot as plt
# import seaborn as sns
# from numba import njit, prange
# import time
from sklearn.model_selection import train_test_split, StratifiedKFold
from tqdm import tqdm
from keras import backend as K
from viziphant.rasterplot import rasterplot

from sklearn.utils import resample
import astropy
import matplotlib.pyplot as plt
from matplotlib.backends.backend_pdf import PdfPages
from matplotlib.ticker import MaxNLocator
import seaborn as sns
from datetime import datetime
from astropy.stats import bootstrap
import sklearn
from instruments.helpers.util import simple_xy_axes, set_font_axes
from instruments.helpers.neural_analysis_helpers import get_word_aligned_raster_squinty, split_cluster_base_on_segment_zola

from helpers.neural_analysis_helpers_zolainter import get_word_aligned_raster, get_word_aligned_raster_squinty
from instruments.helpers.euclidean_classification_minimal_function import classify_sweeps
# Import standard packages
import numpy as np
import matplotlib.pyplot as plt
from scipy import io
from scipy import stats
import pickle



def find_repeating_substring(string):
    length = len(string)
    half_length = length // 2

    # Iterate through possible lengths of the repeating substring
    for i in range(1, half_length + 1):
        substring = string[:i]
        times = length // i

        # Construct the potential repeating substring
        potential_repeat = substring * times

        # Check if the constructed substring matches the original string
        if potential_repeat == string:
            return substring

    return None

def run_cleaning_of_rasters(blocks, datapath):
    clust_ids = [st.annotations['cluster_id'] for st in blocks[0].segments[0].spiketrains if
                 st.annotations['group'] != 'noise']
    for cluster_id in clust_ids:
        new_blocks = split_cluster_base_on_segment_zola(blocks, cluster_id, num_clusters=2)
    with open(datapath / 'new_blocks.pkl', 'wb') as f:
        pickle.dump(new_blocks, f)
    return new_blocks
def target_vs_probe_with_raster(blocks, talker=1,  clust_ids = [], stream = 'BB_3', phydir = 'phy', animal = 'F1702_Zola', brain_area = [], gen_psth = False):

    tarDir = Path(f'E:/rastersms4spikesortinginter/{animal}/npyfiles_andfigs/{phydir}/{stream}/')
    #load the high generalizable clusters, csv file

    saveDir = tarDir
    saveDir.mkdir(exist_ok=True, parents=True)

    binsize = 0.01
    window = [0, 0.6]
    probewords_list = [(1, 1), (2, 2), (3, 3), (4, 4), (5, 5), (6, 6), (7, 7), (8, 8), (9, 9), (10, 10), (11, 11),
                       (12, 12), (13, 13), (14, 14)]
    probewords_list = [ (3, 3), (6,6), (14,14)]


    animal_id_num = animal.split('_')[0]
    # clust_ids = [st.annotations['cluster_id'] for st in blocks[0].segments[0].spiketrains if
    #              st.annotations['group'] != 'noise']
    spiketraindict_means = {}
    for j, cluster_id in enumerate(clust_ids):
        #make a figure of 2 columns and 10 rows
        count = 0
        for idx, probewords in enumerate(probewords_list):
            for pitchshift_option in [True, False]:
                if pitchshift_option:
                    pitchshift_text = 'inter-roved F0'
                else:
                    pitchshift_text = 'control F0'
                fig, ax = plt.subplots()

                raster_target = get_word_aligned_raster_squinty(blocks, cluster_id, word=probewords[0],
                                                                                          pitchshift=pitchshift_option,
                                                                                          correctresp=True,
                                                                                          df_filter=['No Level Cue'])
                raster_target = raster_target.reshape(raster_target.shape[0], )
                if len(raster_target) == 0:
                    print('raster target empty:', cluster_id)
                    continue

                bins = np.arange(window[0], window[1], binsize)

                unique_trials_targ = np.unique(raster_target['trial_num'])
                raster_targ_reshaped = np.empty([len(unique_trials_targ), len(bins) - 1])
                count = 0
                for trial in (unique_trials_targ):
                    raster_targ_reshaped[count, :] = \
                    np.histogram(raster_target['spike_time'][raster_target['trial_num'] == trial], bins=bins,
                                 range=(window[0], window[1]))[0]
                    count += 1

                spiketrains = []
                for trial_id in unique_trials_targ:
                    selected_trials = raster_target[raster_target['trial_num'] == trial_id]
                    spiketrain = neo.SpikeTrain(selected_trials['spike_time'], units='s', t_start=min(selected_trials['spike_time']), t_stop=max(selected_trials['spike_time']))
                    spiketrains.append(spiketrain)

                # try:
                if probewords[0] == 4 and pitchshift_option == False:
                    probeword_text = 'when a'
                    color_option = 'green'
                elif probewords[0] == 4 and pitchshift_option == True:
                    probeword_text = 'when a'
                    color_option = 'lightgreen'

                elif probewords[0] == 1 and pitchshift_option == False:
                    probeword_text = 'instruments'
                    color_option = 'blue'
                elif probewords[0] == 1 and pitchshift_option == True:
                    probeword_text = 'instruments'
                    color_option = 'skyblue'


                elif probewords[0] == 2 and pitchshift_option == False:
                    probeword_text = 'craft'
                    color_option = 'deeppink'
                elif probewords[0] == 2 and pitchshift_option == True:
                    probeword_text = 'craft'
                    color_option = 'pink'

                elif probewords[0] == 3 and pitchshift_option == False:
                    probeword_text = 'in contrast'
                    color_option = 'mediumpurple'
                elif probewords[0] == 3 and pitchshift_option == True:
                    probeword_text = 'in contrast'
                    color_option = 'purple'

                elif probewords[0] == 5 and pitchshift_option == False:
                    probeword_text = 'accurate'
                    color_option = 'black'

                elif probewords[0] == 5 and pitchshift_option == True:
                    probeword_text = 'accurate'
                    color_option = 'grey'
                elif probewords[0] == 6 and pitchshift_option == False:
                    probeword_text = 'pink noise'
                    color_option = 'navy'
                elif probewords[0] == 6 and pitchshift_option == True:
                    probeword_text = 'pink noise'
                    color_option = 'lightblue'

                elif probewords[0] == 7 and pitchshift_option == False:
                    probeword_text = 'of science'
                    color_option = 'coral'
                elif probewords[0] == 7 and pitchshift_option == True:
                    probeword_text = 'of science'
                    color_option = 'orange'


                elif probewords[0] == 8 and pitchshift_option == False:
                    probeword_text = 'rev. instruments'
                    color_option = 'plum'
                elif probewords[0] == 8 and pitchshift_option == True:
                    probeword_text = 'rev. instruments'
                    color_option = 'darkorchid'
                elif probewords[0] == 9 and pitchshift_option == False:
                    probeword_text = 'boats'
                    color_option = 'slategrey'
                elif probewords[0] == 9 and pitchshift_option == True:
                    probeword_text = 'boats'
                    color_option = 'royalblue'

                elif probewords[0] == 10 and pitchshift_option == False:
                    probeword_text = 'today'
                    color_option = 'gold'
                elif probewords[0] == 10 and pitchshift_option == True:
                    probeword_text = 'today'
                    color_option = 'yellow'
                else:
                    probeword_text = 'error'
                    color_option = 'red'
                #if pitchshift plot on the second column
                custom_xlim = (-0.1, 0.6)

                if pitchshift_option:
                    if gen_psth:
                        bin_width = 0.01  # Width of the time bins in seconds
                        time_start = 0  # Start time for the PSTH (in seconds)
                        time_end = 0.6  # End time for the PSTH (in seconds)
                        stimulus_onset = 0.0  # Time of the stimulus onset (relative to the PSTH window)

                        # Calculate PSTH within the specified time range
                        # num_bins = int((time_end - time_start) / bin_width)-1
                        # bins = np.linspace(time_start, time_end, num_bins-1)
                        bins = np.arange(time_start, time_end, bin_width)
                        num_bins = len(bins)-1

                        spike_times = [st.times.magnitude for st in spiketrains]

                        # Flatten spike times and filter within the specified time range
                        spike_times_flat = np.concatenate(spike_times)
                        spike_times_filtered = spike_times_flat[
                            (spike_times_flat >= time_start) & (spike_times_flat <= time_end)]

                        # Compute the histogram within the specified time range
                        hist, _ = np.histogram(spike_times_filtered, bins=bins)

                        # Calculate time axis for plotting within the specified time range
                        time_axis = np.linspace(time_start, time_end, num_bins) + bin_width / 2

                        # Apply smoothing using Gaussian filter\
                        # get the number of trials
                        num_trials = len(unique_trials_targ)
                        # divide the histogram by the number of trials
                        hist_divided_bytrial_num = hist / num_trials
                        # convert to spikes/s
                        hist_rate = hist_divided_bytrial_num / bin_width

                        ax.plot(time_axis, hist_rate, color=color_option, linewidth=2)

                        ax.set_ylabel('spikes/s')
                        ax.set_xlabel('time (s)')
                        dict_key = f'{cluster_id}_{phydir}_{stream}_{probeword_text}_{pitchshift_text}'
                        # spiketraindict[dict_key] = spiketrains
                        spiketraindict_means[dict_key] = hist_rate

                    else:
                        rasterplot(spiketrains, c=color_option, histogram_bins=0, axes=ax, s=0.3)
                        ax.set_ylabel('trial number')
                        ax.set_xlim(custom_xlim)

                    ax.set_title(f'{cluster_id}_{phydir}_{stream}, \n {animal_id_num}, probe word: {probeword_text}, {pitchshift_text}')
                    # ax.text(-0.1, 0.5, probeword_text, horizontalalignment='center',
                    #                 verticalalignment='center', rotation=90, transform=ax.transAxes)
                    plt.savefig(
                        str(saveDir) + f'/PSTH_{gen_psth}_targdist_grid_clusterid_{cluster_id}_probeword_{probewords[0]}_Pitchshift_{pitchshift_option}_{stream}_' + str(
                            cluster_id) + '.png', bbox_inches='tight')
                else:
                    if gen_psth:
                        # get the array of spiketimes
                        bin_width = 0.01  # Width of the time bins in seconds
                        time_start = 0  # Start time for the PSTH (in seconds)
                        time_end = 0.6  # End time for the PSTH (in seconds)
                        stimulus_onset = 0.0  # Time of the stimulus onset (relative to the PSTH window)

                        # Calculate PSTH within the specified time range
                        bins = np.arange(time_start, time_end, bin_width)
                        num_bins = len(bins)-1
                        spike_times = [st.times.magnitude for st in spiketrains]

                        # Flatten spike times and filter within the specified time range
                        spike_times_flat = np.concatenate(spike_times)
                        spike_times_filtered = spike_times_flat[
                            (spike_times_flat >= time_start) & (spike_times_flat <= time_end)]

                        # Compute the histogram within the specified time range
                        hist, _ = np.histogram(spike_times_filtered, bins=bins)

                        # Calculate time axis for plotting within the specified time range
                        time_axis = np.linspace(time_start, time_end, num_bins) + bin_width / 2

                        # Apply smoothing using Gaussian filter\
                        # get the number of trials
                        num_trials = len(unique_trials_targ)
                        # divide the histogram by the number of trials
                        hist_divided_bytrial_num = hist / num_trials
                        # convert to spikes/s
                        hist_rate = hist_divided_bytrial_num / bin_width

                        ax.plot(time_axis, hist_rate, color=color_option, linewidth=2)

                        ax.set_ylabel('spikes/s')
                        ax.set_xlabel('time (s)')
                        dict_key = f'{cluster_id}_{phydir}_{stream}_{probeword_text}_{pitchshift_text}'
                        # spiketraindict[dict_key] = spiketrains
                        spiketraindict_means[dict_key] = hist_rate


                    else:
                        rasterplot(spiketrains, c=color_option, histogram_bins=0, axes=ax, s=0.3)
                        ax.set_ylabel('trial number')

                    ax.set_xlim(custom_xlim)
                    ax.set_title(f'{cluster_id}_{phydir}_{stream},\n {animal_id_num}, probeword: {probeword_text}, {pitchshift_text}')

                    # ax.text(-0.1, 0.5, probeword_text, horizontalalignment='center',
                    #                 verticalalignment='center', rotation=90, transform=ax.transAxes)


                    plt.savefig(
                        str(saveDir) + f'/PSTH_{gen_psth}_targdist_grid_clusterid_{cluster_id}_probeword_{probewords[0]}_Pitchshift_{pitchshift_option}_{stream}_' + str(
                            cluster_id) + '.png', bbox_inches='tight')



                # except Exception:
                #     continue


        # ax[0, 1].set_title('Pitch-shifted F0')
        # ax[0, 0].set_title('Control F0')





    return spiketraindict_means



def generate_rasters(dir):

    datapath_big = Path(f'D:/ms4output_16102023/F1606_Windolene/')
    animal = str(datapath_big).split('\\')[-1]
    datapaths = [x for x in datapath_big.glob('**/mountainsort4/phy//') if x.is_dir()]
    bigoutputfolder = Path(f'E:/rastersms4spikesortinginter/{animal}/npyfiles_dict_highperforming')
    bigoutputfolder.mkdir(exist_ok=True, parents=True)
    all_mean_units_for_animal = []
    for datapath in datapaths:
        stream = str(datapath).split('\\')[-3]
        stream = stream[-4:]
        print(stream)
        folder = str(datapath).split('\\')[-3]
        with open(datapath / 'new_blocks.pkl', 'rb') as f:
            new_blocks = pickle.load(f)


        high_units = pd.read_csv(f'G:/neural_chapter/figures/unit_ids_trained_topgenindex_{animal}.csv')
        # remove trailing steam
        rec_name = folder[:-5]
        #find the unique string
        repeating_substring = find_repeating_substring(rec_name)


        #remove the repeating substring

        # find the units that have the phydir

        max_length = len(rec_name) // 2

        for length in range(1, max_length + 1):
            for i in range(len(rec_name) - length):
                substring = rec_name[i:i + length]
                if rec_name.count(substring) > 1:
                    repeating_substring = substring
                    break

        print(repeating_substring)
        rec_name = repeating_substring
        high_units = high_units[(high_units['recname'] == rec_name) & (high_units['stream'] == stream)]
        clust_ids = high_units['ID'].to_list()
        brain_area = high_units['BrainArea'].to_list()
        clust_ids = [int(str(x).split('.')[0]) for x in clust_ids]
        #just get the unique numbers
        clust_ids = list(set(clust_ids))

        # if clust_ids == []:
        #     print('no units found')
        #     continue
        for talker in [1]:
            spiketraindict_means = target_vs_probe_with_raster(new_blocks,clust_ids = clust_ids, talker=talker, stream = stream, phydir=repeating_substring, animal = animal, brain_area = brain_area, gen_psth  = True)
            all_mean_units_for_animal.append(spiketraindict_means)
            #append the spiketraindict_means to a list
            #save the list as a pickle
    with open(bigoutputfolder / 'spiketraindict_means.pkl', 'wb') as f:
        pickle.dump(all_mean_units_for_animal, f)

            # target_vs_probe_with_raster(new_blocks,clust_ids = clust_ids, talker=talker, stream = stream, phydir=repeating_substring, animal = animal, brain_area = brain_area, gen_psth=True)




def main():

    directories = ['zola_2022']  # , 'Trifle_July_2022']
    for dir in directories:
        generate_rasters(dir)


if __name__ == '__main__':
    main()