Tutorial 08 - MedPC behavior

Goals

• Implement a behavioral analysis for a Pavlovian task measured in epochs

Compute the duration in food cup for rewarded and unrewarded trials

This data is from a Pavlovian biconditional discrimination experiment where two lights (steady, flashing) and two sounds (tone, white noise) were presented for 10 seconds at a time. A trial consisted of one light presentation followed by a 5 second gap, then one sound presentation. There were four different trial types (AX-, AY+, BX+, BY-), such that each individual cue is equally presented with and without food reward. A successful discrimination in this task is when AY+ and BX+ trials have a greater response than AX- and BY- trials.

# Import necessary packages
%matplotlib inline
import os
import numpy as np
import nept
import matplotlib.pyplot as plt
import seaborn as sns

# Define where your data folder is located
data_path = os.path.join(os.path.abspath('.'), 'data')
data_folder = os.path.join(data_path, 'biconditional-tutorial')

# Let's look at Rat 5
rats = ['5']
group1 = ['1', '3', '5', '7']
group2 = ['2', '4', '6', '8']

# Get the sessions. MedPC data is stored in a file that is is in the format !YYYY-MM-DD.
# The a and b indicate morning and afternoon sessions.
sessions = []
for file in sorted(os.listdir(data_folder)):
    if file[0] == '!':
        sessions.append(os.path.join(data_folder, file))

# Find out the number of sessions
n_sessions = len(sessions)
print('Number of sessions:', n_sessions)

Number of sessions: 40

# Write a function that dictates the maping between the medpc values and the experimental conditions
def assign_medpc_label(data):
    """Assigns events to proper labels.

    Parameters
    ----------
    data: dict

    Returns
    -------
    rats_data: dict
        With mags, pellets, lights1, lights2, sounds1, sounds2, trial1, trial2, trial3, trial4 as keys.
        Each contains nept.Epoch objects

    """
    mag_start = np.array(data[1])
    mag_end = np.array(data[2])
    if len(mag_start) > len(mag_end):
        mag_start = np.array(data[1][:-1])
    pel_start = np.array(data[3])
    pel_end = pel_start + 1
    light1_start = np.array(data[4])
    light1_end = np.array(data[5])
    light2_start = np.array(data[6])
    light2_end = np.array(data[7])
    sound1_start = np.array(data[8])
    sound1_end = np.array(data[9])
    sound2_start = np.array(data[10])
    sound2_end = np.array(data[11])

    rats_data = dict()
    rats_data['mags'] = nept.Epoch(mag_start, mag_end-mag_start)
    rats_data['pellets'] = nept.Epoch(pel_start, pel_end-pel_start)
    rats_data['lights1'] = nept.Epoch(light1_start, light1_end-light1_start)
    rats_data['lights2'] = nept.Epoch(light2_start, light2_end-light2_start)
    rats_data['sounds1'] = nept.Epoch(sound1_start, sound1_end-sound1_start)
    rats_data['sounds2'] = nept.Epoch(sound2_start, sound2_end-sound2_start)

    return rats_data

# Set up the data in a dictionary, which is most useful when handling multiple rats
data = dict()
for rat in rats:
    data[rat] = nept.Rat(rat, group1, group2)

# Gather the data from all the sessions and add them to the dictionary
for session in sessions:
    rats_data = nept.load_medpc(os.path.join(data_folder, session), assign_medpc_label)

    for rat in rats:
        data[rat].add_session_medpc(**rats_data[rat])

# Make a dataframe from the dictionary
df = nept.combine_rats(data, rats, n_sessions, only_sound=False)

# Take a look at the first few entries of the dataframe.
# Notice we're computing more than only the food cup duration here.
# We can filter it to only the values of interest in our plotting.
df.head()

	condition	cue	measure	rat	rewarded	session	trial	trial_type	value
0	light 1	light	durations	5	light unrewarded	1	5, 0	1	0.00
1	light 1	light	numbers	5	light unrewarded	1	5, 0	1	0.00
2	light 1	light	latency	5	light unrewarded	1	5, 0	1	10.00
3	light 1	light	responses	5	light unrewarded	1	5, 0	1	0.00
4	sound 1	sound	durations	5	sound unrewarded	1	5, 1	1	0.12

# Plot the duration in food cup behavior
sns.set_style("white")

def plot_measure(measure):
    """Plots the measure overtime for one rat.

    Parameters
    ----------
    measure: str
        Must be 'durations', 'numbers', 'latency', or 'responses'.

    """
    # Filter the dataframe by rat
    rat_idx = np.zeros(len(df), dtype=bool)

    for rat in rats:
        rat_idx = rat_idx | (df['rat'] == rat)
    rats_df = df[rat_idx]

    # Filter the dataframe by measure, here let's look at the duration of food cup entries
    duration = df.loc[df.measure == measure]

    # Plot using seaborn since seaborn and pandas (the dataframe) work well together
    colours = ["#9970ab", "#d6604d", "#1b7837", "#2166ac"]
    ax = sns.tsplot(data=duration, time="session", unit="trial", condition="rewarded", value="value",
                    err_style="ci_band", ci=68, color=colours)

    # Set plotting parameters
    legend_dist = 1.

    if measure == 'durations':
        ax.set(xlabel='Session', ylabel='Duration in food cup (s)')
    elif measure == 'numbers':
        ax.set(xlabel='Session', ylabel='Number of food cup entries')
    elif measure == 'latency':
        ax.set(xlabel='Session', ylabel='Latency to first entry (s)')
    elif measure == 'responses':
        ax.set(xlabel='Session', ylabel='Percent responses')

    sns.despine()
    plt.tight_layout()
    plt.legend(bbox_to_anchor=(legend_dist, 1.))
    plt.title('Rat 5 duration in food cup by outcome')

    # Show the plot
    plt.show()

plot_measure('durations')

plot_measure('numbers')