-
Notifications
You must be signed in to change notification settings - Fork 3
/
gen_figure3_eyetracking.py
238 lines (166 loc) · 9.43 KB
/
gen_figure3_eyetracking.py
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
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Assess and plot eye tracking data quality recorded during the EMU sessions
"""
import os
import numpy as np
import pandas as pd
from pynwb import NWBHDF5IO
from tqdm import tqdm
from glob import glob
from ephys_utills import get_et_timebins, compute_pixelperDVA, et_heatmap
import matplotlib.pyplot as plt
import argparse
# Set Matplotlib SVG font type
plt.rcParams['svg.fonttype'] = 'none'
SEC2MSEC = 1000.
def main(nwb_input_dir):
# ----- Basic metadata info about the video stimulus -----
frame_width, frame_height, vid_fps, nframes = 640, 480, 25.0, 11971
frame_duration_msec = SEC2MSEC / vid_fps
framesize = [frame_width, frame_height]
# define a duration to timebin the ET data
timebin_sec = 1.0 # define here as sec. used for saving results
timebin_msec = timebin_sec * SEC2MSEC # millisec.
nbins = np.round(frame_duration_msec * nframes / timebin_msec).astype(int)
nwb_session_files = sorted(glob(os.path.join(nwb_input_dir, 'sub-*/*.nwb')))
# ----- Read ET (gaze) data from the NWB files -----
etdata_ses = []
pixel_dva_ses = []
missingdata_ratio = []
session_ids_et = []
for session_ii in nwb_session_files:
print(f'processing {os.path.basename(session_ii)}...')
# Open the NWB file and read its content
with NWBHDF5IO(session_ii,'r') as nwb_io:
nwbfile = nwb_io.read()
session_ids_et.append(nwbfile.identifier)
trials_df = nwbfile.trials.to_dataframe()
enc_start_time = trials_df[trials_df['stim_phase']=='encoding']['start_time'].values[0]
enc_stop_time = trials_df[trials_df['stim_phase']=='encoding']['stop_time'].values[0]
gaze_data = nwbfile.processing['behavior']['EyeTracking']['SpatialSeries']
gaze_xy = np.asarray(gaze_data.data)
if gaze_data.rate is None:
gaze_time = gaze_data.timestamps
else:
gaze_time = np.arange(0,len(gaze_xy))/(gaze_data.rate) + gaze_data.starting_time
gaze_encoding = np.logical_and(gaze_time >= enc_start_time,
gaze_time <= enc_stop_time)
gaze_df = pd.DataFrame(data=np.c_[gaze_time[gaze_encoding],gaze_xy[gaze_encoding,:]],
columns=['RecTime','GazeX','GazeY'])
# get video display info and scale gaze to stimulus size
display_info_raw = gaze_data.comments
screen_wh, display_wh, display_area_i = display_info_raw.split('::')
screen_w, screen_h = list(map(float,screen_wh.split('=')[1].split(',')))
display_w, display_h = list(map(float,display_wh.split('=')[1].split(',')))
display_area = list(map(float,display_area_i.split('=')[1].split(',')))
_, pixel_dva_mean = compute_pixelperDVA([screen_w,screen_h])
pixel_dva_ses.append(pixel_dva_mean)
scale_dx = frame_width / display_w
scale_dy = frame_height / display_h
gaze_df['GazeX'] = (gaze_df['GazeX'] - display_area[0])*scale_dx
gaze_df['GazeY'] = (gaze_df['GazeY'] - display_area[1])*scale_dy
problem_inds_x = np.logical_not(gaze_df['GazeX'].astype(float).between(0,frame_width,
inclusive='left'))
problem_inds_y = np.logical_not(gaze_df['GazeY'].astype(float).between(0,frame_height,
inclusive='left'))
problem_inds = np.logical_or(problem_inds_x, problem_inds_y)
gaze_df.loc[problem_inds,['GazeX','GazeY']] = np.NaN
# Downsample gaze data to the video frame rate
et_xy_binned = get_et_timebins(gaze_df, timebin_msec, do_op=None,
fix_length=True, nbins=nbins, keep_timebin_index=False,
)
et_xy_binned_v = np.vstack(et_xy_binned)
missingdata_ratio.append(np.isnan(et_xy_binned_v).any(axis=1).sum() / et_xy_binned_v.shape[0])
et_xy_binned = np.asarray(et_xy_binned, dtype=object)
etdata_ses.append(et_xy_binned)
# --- data to be used for ET analysis ---
session_ids_et = np.asarray(session_ids_et)
re_idx = np.argsort(session_ids_et) # resort subject IDs, all in _r1, _r2 order
session_ids_et = session_ids_et[re_idx]
etdata_ses = np.asarray(etdata_ses, dtype=object)[re_idx,:]
pixel_dva_ses = np.asarray(pixel_dva_ses)[re_idx]
missingdata_ratio = np.asarray(missingdata_ratio)[re_idx]
# ----- Compute heatmap correlations between subjects -----
sigma = np.mean(pixel_dva_ses) # Standard deviation for Gaussian kernel. Equal for both axes.
hp_down_factor = float(5.0) # should be float. Downsample heatmaps for a quick analysis
nan_ratio = 0.50
# remove subjects with missing data more than half of all recording time
remove_subjs_indx = missingdata_ratio>=nan_ratio
print(session_ids_et[remove_subjs_indx])
session_ids_et = session_ids_et[~remove_subjs_indx]
etdata_ses = etdata_ses[~remove_subjs_indx]
pixel_dva_ses = pixel_dva_ses[~remove_subjs_indx]
missingdata_ratio = missingdata_ratio[~remove_subjs_indx]
unique_subjs = np.unique([ sii.split('_')[0] for sii in session_ids_et ])
subj2agg_corrs = np.zeros((len(session_ids_et), nbins))
for tii in tqdm(range(nbins)):
etdata_tbin = etdata_ses[:,tii,...]
for sub_ii in unique_subjs:
subj_this_bool = [ True if sub_ii in sii else False for sii in session_ids_et ]
subj_other_bool = np.logical_not(subj_this_bool)
# reference heatmap from all other subjects:
agg_data = np.vstack(etdata_tbin[subj_other_bool,...])
agg_hmap = et_heatmap(agg_data, framesize, sigma, hp_down_factor, get_full=False)
subj_this = session_ids_et[subj_this_bool]
for sub_jj in subj_this:
sub_idx = session_ids_et.tolist().index(sub_jj)
subj_hmap = et_heatmap(etdata_tbin[sub_idx], framesize, sigma, #pixel_dva_ses[sub_idx],
hp_down_factor, get_full=False, nan_ratio=nan_ratio)
if subj_hmap is not None:
subj2agg_corrs[sub_idx, tii] = np.corrcoef(agg_hmap.ravel(), subj_hmap.ravel())[0,1]
else:
subj2agg_corrs[sub_idx, tii] = np.NaN
# report overall ET correlations
subj2agg_corrs_z = np.arctanh(subj2agg_corrs)
mean_hmap_corrs = np.tanh( np.nanmean(subj2agg_corrs_z , 1) )
mean_hmap_corrs_low = np.tanh( np.nanpercentile(subj2agg_corrs_z, 16, 1) )
mean_hmap_corrs_up = np.tanh( np.nanpercentile(subj2agg_corrs_z, 84, 1) )
print(np.tanh(np.mean(np.nanmean(subj2agg_corrs_z, 1))),
np.tanh(np.std(np.nanmean(subj2agg_corrs_z, 1))) )
# ----- Plot obtained gaze metrics -----
fcol = '#009ccc'
fcol_txt = '#008db8'
session_ids_et_txt = [ sii.lower() for sii in session_ids_et ]
pos = np.arange(len(session_ids_et))
error_config =dict(ecolor ='#006f91', linewidth=1, capsize=2, capthick=1.)
fig, ax = plt.subplots(1, 1, figsize=(6, 2.35))
ax.bar(pos, mean_hmap_corrs,
yerr=[ mean_hmap_corrs-mean_hmap_corrs_low, mean_hmap_corrs_up-mean_hmap_corrs ] ,
color=fcol, error_kw=error_config, label='')
ax.set_ylim([0,1])
ax.set_ylabel('Average gaze heatmap\n correlation (r)', fontsize=8, color=fcol_txt)
ax.tick_params(axis='y', which='major', labelsize=7, length=3, pad=1,
labelcolor=fcol_txt, color=fcol_txt
)
ax.spines['top'].set_visible(False)
ax.get_xaxis().tick_bottom()
ax.margins(0.013)
fcol2 = '#787878'
ax2 = ax.twinx()
ax2.bar(pos, missingdata_ratio*100, color=fcol2, width=0.4)
ax2.set_ylim([0,100])
ax2.set_ylabel('Percentage of missing\n gaze data (%)', fontsize=8, color=fcol2)
ax2.tick_params(axis='y', which='major', labelsize=7, length=3, pad=1,
labelcolor=fcol2, color=fcol2,
)
ax2.spines['top'].set_visible(False)
ax2.get_xaxis().tick_bottom()
ax2.margins(0.013)
ax2.spines['right'].set_color(fcol2)
ax2.spines['left'].set_color(fcol_txt)
ax.set_xticks(pos)
ax.set_xticklabels(session_ids_et_txt, fontsize=7, rotation=80, ha='center')
ax.tick_params(axis='x', which='major', length=3, pad=0)
plt.tight_layout()
fig.savefig("gaze_group_mean-hmap.png", dpi=600, bbox_inches='tight')
fig.savefig("gaze_group_mean-hmap.svg", format='svg', dpi=600, bbox_inches='tight')
if __name__ == '__main__':
parser = argparse.ArgumentParser(description="Load NWB files to plot eye tracking data quality.")
parser.add_argument('--nwb_input_dir', type=str, required=True, help='Directory containing NWB files.')
args = parser.parse_args()
main(args.nwb_input_dir)
'''
python gen_figure3_eyetracking.py --nwb_input_dir /path/to/nwb_files/
'''