openneuro/ds003604
Watch
3
0
Fork
You've already forked ds003604
 0
Code Pull requests Packages Activity
ds003604/code/beh_process.py
Brain Development Lab 4f9dac0c5a [DATALAD] Recorded changes
2021-04-08 18:48:11 +00:00

197 lines
12 KiB (Stored with Git Annex)
Python
Raw Permalink Blame History

'''
Purpose
How to Run
Marisa Lytle
Feb 2020
'''
import pandas as pd
import numpy as np
def deidentify(data,info):
#Read in birthdate data
bday_file = "./Birthdates.csv"
birthdates = pd.read_csv(bday_file)
birthdates['birthdate'] = pd.to_datetime(birthdates['birthdate'], format="%m/%d/%Y")
#Convert session date to datetime object
data['SessionDate2'] = pd.to_datetime(data['SessionDate'], format="%m-%d-%Y")
#Merge in birthdates
data = pd.merge(data, birthdates, on='subject', how='left')
#Calculate age in days
data['age'] = data['SessionDate2'] - data['birthdate']
data['age'] = data['age'].dt.days
data['ShiftedSessionDate'] = data['SessionDate2'] + pd.DateOffset(days=info['shift']) - pd.DateOffset(years=200)
data = data.drop(['SessionDate','SessionDate2', 'birthdate', 'Clock.Information', 'SessionStartDateTimeUtc'], axis=1)
return data
def onset_calc(run_data):
#Calculate average onsets to use for subjects that cannot be calculated due to missing task or trial onsets
if run_data['task_onset'].isnull().values.any() or run_data['trial_onset'].isnull().values.any():
run_data['onset'] = 0
elif run_data['task_onset'].notnull().values.any() and run_data['trial_onset'].notnull().values.any():
run_data['onset'] = (run_data['trial_onset'] - run_data['task_onset']).round(4)
return run_data
def clean_col(data):
if 'Present.OnsetTime' not in data.columns:
data = data.rename(columns={"Present.RESP": "window1_response", "Present.RT": "window1_RT",
"Response.RESP": "window2_response", "Response.RT": "window2_RT", "SoundFile": "stim_file",
"Condition": "trial_type", "Subject": "subject", "Response.Duration": "window2_dur", "Correct": "correct"})
data['task_onset1'] = 0
data['task_onset2'] = 0
data['trial_onset'] = 0
elif 'Blank.OnsetTime' not in data.columns:
data = data.rename(columns={"Blank1.OnsetTime": "task_onset2",
"Present.OnsetTime": "trial_onset", "Present.RESP": "window1_response", "Present.RT": "window1_RT",
"Response.RESP": "window2_response", "Response.RT": "window2_RT", "SoundFile": "stim_file",
"Condition": "trial_type", "Subject": "subject", "Response.Duration": "window2_dur", "Correct": "correct"})
data['task_onset1'] = 0
elif 'Blank1.OnsetTime' not in data.columns:
data = data.rename(columns={"Blank.OnsetTime": "task_onset1",
"Present.OnsetTime": "trial_onset", "Present.RESP": "window1_response", "Present.RT": "window1_RT",
"Response.RESP": "window2_response", "Response.RT": "window2_RT", "SoundFile": "stim_file",
"Condition": "trial_type", "Subject": "subject", "Response.Duration": "window2_dur", "Correct": "correct"})
data['task_onset2'] = 0
else:
data = data.rename(columns={"Blank.OnsetTime": "task_onset1","Blank1.OnsetTime": "task_onset2",
"Present.OnsetTime": "trial_onset", "Present.RESP": "window1_response", "Present.RT": "window1_RT",
"Response.RESP": "window2_response", "Response.RT": "window2_RT", "SoundFile": "stim_file",
"Condition": "trial_type", "Subject": "subject", "Response.Duration": "window2_dur", "Correct": "correct"})
#initialize task onset column
data['task_onset'] = 0
#merge the two different task onset columns together as Gram1 uses task_onset1 and others use task_onset2
for x in range(len(data['subject'].values)):
if data['task_onset1'].iloc[x] > 0:
data.loc[x,'task_onset'] = data['task_onset1'].iloc[x]
elif data['task_onset2'].iloc[x] > 0:
data.loc[x,'task_onset'] = data['task_onset2'].iloc[x]
else:
data.loc[x,'task_onset'] = np.nan
#Convert ms to sec
data[['task_onset', 'trial_onset', 'window1_RT', 'window2_RT', 'window2_dur']] \
= data[['task_onset', 'trial_onset', 'window1_RT', 'window2_RT', 'window2_dur']].applymap(lambda x: x / 1000)
#Take off extra characters from Experiment name and stimulus names
data['ExperimentName'] = data['ExperimentName'].map(lambda x: x.lstrip('9_'))
data['ExperimentName'] = data['ExperimentName'].map(lambda x: x.rstrip('_font'))
data['ExperimentName'] = data['ExperimentName'].map(lambda x: x.rstrip('_onsets'))
data['ExperimentName'] = data['ExperimentName'].map(lambda x: x.rstrip('_test'))
data['ExperimentName'] = data['ExperimentName'].map(lambda x: x.rstrip(' - Final'))
data['ExperimentName'] = data['ExperimentName'].map(lambda x: x.rstrip('_gray'))
data['stim_file'] = data['stim_file'].map(lambda x: x.lstrip('Stimuli/'))
data['window1_response_raw'] = data['window1_response']
#correct reaction times listed as 0 to nan for non responses
for x in range(len(data['window1_RT'].values)):
if np.isnan(data['window1_response'].iloc[x]):
data.loc[x,'window1_RT'] = np.nan
if np.isnan(data['window2_response'].iloc[x]):
data.loc[x,'window2_RT'] = np.nan
if np.isnan(data['window1_RT'].iloc[x]):
data.loc[x,'window1_response'] = np.nan
return data
def calc_rt_acc(run_data):
control_trials = ['G_GC', 'P_SC', 'SP_Co', 'S_SC']
#Initialize calculated columns
run_data['calculated_response'] = 0
run_data['calculated_RT'] = 0
run_data['calculated_accuracy'] = 0
run_data['duration'] = run_data["window2_dur"] + 1.75
#Do the RT and ACC calculations following rules listed in ELP_TaskAccRT.docx
for trial in range(len(run_data['subject'].values)):
if trial == (len(run_data['subject'].values)-1):
if pd.notnull(run_data['window2_response'].iloc[trial]) and run_data['window2_RT'].iloc[trial] > 0.25:
run_data.loc[trial,'calculated_response'] = run_data['window2_response'].iloc[trial]
run_data.loc[trial,'calculated_RT'] = run_data['window2_RT'].iloc[trial] + 0.75
elif pd.isnull(run_data['window2_response'].iloc[trial]) or run_data['window2_RT'].iloc[trial] < 0.25:
run_data.loc[trial,'calculated_response'] = np.nan
run_data.loc[trial,'calculated_RT'] = np.nan
elif run_data['trial_type'].iloc[trial] not in control_trials:
if pd.notnull(run_data['window2_response'].iloc[trial]) and run_data['window2_RT'].iloc[trial] > 0.25 :
run_data.loc[trial,'calculated_response'] = run_data['window2_response'].iloc[trial]
run_data.loc[trial,'calculated_RT'] = run_data['window2_RT'].iloc[trial] + 0.75
elif (pd.isnull(run_data['window2_response'].iloc[trial]) or run_data['window2_RT'].iloc[trial] < 0.25) \
and (run_data['trial_type'].iloc[trial+1] in control_trials):
run_data.loc[trial,'calculated_response'] = np.nan
run_data.loc[trial,'calculated_RT'] = np.nan
elif (pd.isnull(run_data['window2_response'].iloc[trial]) or run_data['window2_RT'].iloc[trial] < 0.25)\
and (run_data['trial_type'].iloc[trial+1] not in control_trials) and pd.notnull(run_data['window1_response'].iloc[trial+1]):
run_data.loc[trial,'calculated_response'] = run_data['window1_response'].iloc[trial+1]
run_data.loc[trial,'calculated_RT'] = run_data['window1_RT'].iloc[trial+1] + run_data['duration'].iloc[trial]
elif (pd.isnull(run_data['window2_response'].iloc[trial]) or run_data['window2_RT'].iloc[trial] < 0.25)\
and (run_data['trial_type'].iloc[trial+1] not in control_trials) and pd.isnull(run_data['window1_response'].iloc[trial+1])\
and run_data['window2_RT'].iloc[trial+1] < 0.25 and pd.notnull(run_data['window2_response'].iloc[trial+1]):
run_data.loc[trial,'calculated_response'] = run_data['window2_response'].iloc[trial+1]
run_data.loc[trial,'calculated_RT'] = run_data['duration'].iloc[trial] + 0.75 + run_data['window2_RT'].iloc[trial+1]
else:
run_data.loc[trial,'calculated_response'] = np.nan
run_data.loc[trial,'calculated_RT'] = np.nan
elif run_data['trial_type'].iloc[trial] in control_trials:
if pd.notnull(run_data['window2_response'].iloc[trial]):
run_data.loc[trial,'calculated_response'] = run_data['window2_response'].iloc[trial]
run_data.loc[trial,'calculated_RT'] = run_data['window2_RT'].iloc[trial] + 0.75
elif pd.isnull(run_data['window2_response'].iloc[trial]) and pd.notnull(run_data['window1_response'].iloc[trial]):
run_data.loc[trial,'calculated_response'] = run_data['window1_response'].iloc[trial]
run_data.loc[trial,'calculated_RT'] = run_data['window1_RT'].iloc[trial]
elif run_data['trial_type'].iloc[trial+1] in control_trials:
run_data.loc[trial,'calculated_response'] = np.nan
run_data.loc[trial,'calculated_RT'] = np.nan
elif pd.isnull(run_data['window2_response'].iloc[trial]) \
and (run_data['trial_type'].iloc[trial+1] not in control_trials) and (pd.notnull(run_data['window1_response'].iloc[trial+1])):
run_data.loc[trial,'calculated_response'] = run_data['window1_response'].iloc[trial+1]
run_data.loc[trial,'calculated_RT'] = run_data['window1_RT'].iloc[trial+1] + run_data['duration'].iloc[trial]
elif pd.isnull(run_data['window2_response'].iloc[trial]) \
and (run_data['trial_type'].iloc[trial+1] not in control_trials) and (pd.isnull(run_data['window1_response'].iloc[trial+1]))\
and (run_data['window2_RT'].iloc[trial+1] < 0.25 and pd.notnull(run_data['window2_response'].iloc[trial+1])):
run_data.loc[trial,'calculated_response'] = run_data['window2_response'].iloc[trial+1]
run_data.loc[trial,'calculated_RT'] = run_data['duration'].iloc[trial] + 0.75 + run_data['window2_RT'].iloc[trial+1]
else:
run_data.loc[trial,'calculated_response'] = np.nan
run_data.loc[trial,'calculated_RT'] = np.nan
#Determine if calculated response is correct and save to calculated_accuracy column
for ans in range(len(run_data['correct'].values)):
if run_data['correct'].iloc[ans] == run_data['calculated_response'].iloc[ans]:
run_data.loc[ans,'calculated_accuracy'] = 1
elif run_data['correct'].iloc[ans] != run_data['calculated_response'].iloc[ans]:
run_data.loc[ans,'calculated_accuracy'] = 0
#Calculate accuracy, RT, and duration column
run_data['duration'] = run_data['duration'].round(4)
#all nan data to 'n/a'
run_data['calculated_RT'] = run_data['calculated_RT'].round(4)
for col in list(run_data):
run_data[col] = run_data[col].replace(np.nan, 'n/a')
run_data['window1_response'] = run_data['window1_response_raw']
run_data = run_data.drop(['unique_time'], axis=1)
events = run_data[['onset', 'duration', 'trial_type', 'correct', 'window1_response', 'window1_RT', 'window2_response', 'window2_RT', 'stim_file', 'calculated_accuracy',
'calculated_RT']]
return run_data, events
View git blame Copy permalink