DBSpulse/analysis.py at master · nwilming/DBSpulse · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
import numpy as np
import pandas as pd
from scipy.interpolate import interp1d
from os import path
import cPickle
import pylab as plt
import matplotlib
import seaborn as sns
sns.set_style('ticks')


def get_dataset(dir):
    a = cPickle.load(open(path.join(dir,'pupil_data')))
    df = pd.concat([to_df(k) for k in a['gaze_positions']])
    df = regular(df)
    if not 'DBSTrigger' in a.keys():
        return df
    triggers = pd.to_datetime(a['DBSTrigger'], unit='s')
    triggers = pd.DatetimeIndex([df.index.asof(x) for x in triggers])
    return df, triggers


def to_df(datum):
    d = {
         'xc':datum['norm_pos'][0],
         'yc':datum['norm_pos'][1],
         'confidence':datum['confidence']
         }
    for eye in datum['base']:
        id = str(eye['id'])
        d.update({
             'x'+id:eye['norm_pos'][0],
             'y'+id:eye['norm_pos'][1],
             'confidence'+id:eye['confidence'],
             'diameter'+id:eye['diameter']
             })
    for key in datum.keys():
        if key.startswith('realtime gaze'):
            data = datum[key]
            d.update({key+'x':data[0], key+'y':data[0]})

    return pd.DataFrame(d, index=[datum['timestamp']])


def interp(x, y, target):
    f = interp1d(x.values.astype(int), y)
    target = target[target.values.astype(int)>min(x.values.astype(int))]
    return pd.DataFrame({y.name:f(target.values.astype(int))}, index=target)


def regular(df):
    dt = pd.to_datetime(df.index.values, unit='s')
    df = df.set_index(dt)
    target = df.resample('16ms').mean().index
    return pd.concat([interp(dt, df[c], target) for c in df.columns], axis=1)


def time_locked(data, time_points, span):
    data = pd.DataFrame(data)
    return pd.concat(
        [time_lock_series(data[d], time_points, span).set_index(['Time', 'Trial']) for d in data], axis=1)

def time_lock_series(data, time_points, span):
    '''
    data should be a Series!
    '''

    data = pd.Series(data)

    def reindex(x, trig):
        x.index = x.index-trig
        return x

    beg, end = pd.Timedelta(span[0], unit='s'), pd.Timedelta(span[1], unit='s')
    oo = pd.concat(
                    [reindex(data.loc[trig-beg:trig+end], trig) for trig in time_points],
                    axis=1)
    oo.columns = np.arange(oo.shape[1])
    oo.columns.name='Trial'
    oo.index.name='Time'
    oo = oo.set_index(np.linspace(span[0], span[1], oo.shape[0]))
    oo = oo.stack().reset_index()
    oo.columns = ['Time', 'Trial'] + [data.name]
    return oo


def baseline(df, time=(-1, 0)):
    return df-df.loc[[slice(-0.5, 0), None]].mean()


def timelocked_analysis(df, trigs, field='diameter', bad_trigs=None):
    if bad_trigs is None:
        bad_trigs = array([False]*len(trigs))
    df = df.query('confidence>0.8')
    df.loc[:, field] = (df[field]-df[field].mean())/df[field].std()
    plt.figure(figsize=(10, 5))
    gs = matplotlib.gridspec.GridSpec(6, 1)
    plt.subplot(gs[1:, :])

    oo = time_locked(df, trigs[~bad_trigs], [-2.5, 5])
    oo = oo.groupby(level='Trial').apply(baseline)
    sns.tsplot(oo.reset_index(), time='Time', unit='Trial', value=field, estimator=plt.nanmean)
    plt.axvline(0, color='k')
    plt.ylim([-1, 1])
    sns.despine()
    plt.subplot(gs[0, :])
    plt.plot(df[field])
    plt.plot([df.index[0], df.index[0] + pd.Timedelta(10, unit='s')], [-2, -2], 'r')
    for i, l in enumerate(trigs):
        if bad_trigs[i]:
            plt.axvline(l, color='k', alpha=0.5)
        else:
            plt.axvline(l, color='r', alpha=0.5)
    plt.xticks([])
    plt.yticks([])
    sns.despine()


def foo():
    for i, ds in enumerate(['%03i'%i for i in arange(4, 10)]):
        df, trigs = analysis.get_dataset('/Users/nwilming/recordings/2016_06_01/%s'%ds)
        df.diameter = (df.diameter-df.diameter.mean())/df.diameter.std()
        plt.figure(figsize=(10, 5))
        gs = matplotlib.gridspec.GridSpec(6, 1)
        plt.subplot(gs[1:, :])
        oo = analysis.time_locked(df, trigs, [-2.5, 5])
        sns.tsplot(oo.reset_index(), time='Time', unit='Trial', value='diameter', estimator=nanmean)
        plt.axvline(0, color='k')
        plt.ylim([-1, 1])
        sns.despine()
        plt.subplot(gs[0, :])
        plt.plot(df[field])
        plt.plot([df.index[0], df.index[0] + pd.Timedelta(7.5, unit='s')], [-2, -2], 'r')
        for l in trigs:
            plt/axvline(l, color='k', alpha=0.5)
        xticks([])
        yticks([])
        sns.despine()
        savefig('/Users/nwilming/u/DBSpulse/plots/pilot_1_rec_%s.pdf'%ds, bbox_inches='tight')


def foo2():
    for d in ds[3:]:
        t = d.split('/')[-1]
        try:
            df, trigs = analysis.get_dataset(d)
            bad_trigs = array([False]*len(trigs))
            if t == '004':
                bad_trigs[:10] = True
            elif t == '006':
                bad_trigs[4:6] = True
            else:
                bad_trigs[0] = True
            analysis.timelocked_analysis(df, trigs, field='diameter0', bad_trigs=bad_trigs)
            title('Dataset: %s'%t)
            savefig('/Users/nwilming/Dropbox/Pupillometry/2016_11_25/%s.pdf'%t)
        except ValueError:
            print 'No triggers for ', d