Source code for BrainNetworksInPython.scripts.useful_functions

#!/usr/bin/env python


def read_in_df(data_file, aparc_names):
    '''
    A very useful command for NSPN behavmerge data frames
    Beware though - this is quite specific and there are
    a few versions floating around! Be careful
    '''
    import pandas as pd
    import numpy as np
    import os

    # Read in the data file
    df = pd.read_csv(data_file, sep=',')

    # Only keep the first scan!
    df = df.loc[df.occ==0, :]

    # Strip "thickness" or "thicknessstd" from the column
    # names so they match with the aparc_names names
    data_cols = [ x.replace('_{}'.format('thicknessstd'), '') for x in df.columns ]
    df.columns = data_cols
    data_cols = [ x.replace('_{}'.format('thickness'), '') for x in df.columns ]
    df.columns = data_cols

    # Define a few variables you'll want in the data frame
    df['ones'] = df['age_scan'] * 0 + 1
    df['age'] = df['age_scan']

    df['Global'] = df[aparc_names].mean(axis=1)
    df['Global_std'] = df[aparc_names].mean(axis=1)

    # If there is a corresponding standard deviation
    # file then read in the standard deviation
    if 'mean' in data_file:
        std_data_file = data_file.replace('mean', 'std')
    else:
        std_data_file = data_file.replace('thickness', 'thicknessstd')

    if os.path.isfile(std_data_file):

        # Repeating the steps really
        # Read in the file
        df_std = pd.read_csv(std_data_file, sep=',')
        # Only keep the first occ
        df_std = df_std.loc[df_std.occ==0, :]
        # Change the names so they match up
        data_cols = [ x.replace('_{}'.format('thicknessstd'), '') for x in df_std.columns ]
        df_std.columns = data_cols
        data_cols = [ x.replace('_{}'.format('thickness'), '') for x in df_std.columns ]
        df_std.columns = data_cols

        # Now write the std across all aparc names into the original data frame
        # by averaging the variances
        df['Global_std'] = np.sqrt(np.average(df_std[aparc_names]**2, axis=1))

    # Convert the values to floats
    df[aparc_names] = df[aparc_names].astype('float')

    # If this is an MT, R2s, synthetic, MD, L1 or L23 file
    # then you have to divide the values by 1000
    # However there have been problems here in the past with
    # mixing multiplied with non-multiplied values
    # so we'll actually just check for values greater than a
    # reasonable maximum and divide those ones.
    cols_list = aparc_names+['Global']+['Global_std']
    if 'MT' in os.path.basename(data_file):
        df.loc[df['Global']>50, cols_list] = df.loc[df['Global']>50, cols_list]/1000.0
    if 'synthetic' in os.path.basename(data_file):
        df.loc[df['Global']>50, cols_list] = df.loc[df['Global']>50, cols_list]/1000.0
    if 'R2s' in os.path.basename(data_file):
        df.loc[df['Global']>1, cols_list] = df.loc[df['Global']>1, cols_list]/1000.0
    if 'L1' in os.path.basename(data_file):
        df.loc[df['Global']>0.01, cols_list] = df.loc[df['Global']>0.01, cols_list]/1000.0
    if 'L23' in os.path.basename(data_file):
        df.loc[df['Global']>0.01, cols_list] = df.loc[df['Global']>0.01, cols_list]/1000.0
    if 'MD' in os.path.basename(data_file):
        df.loc[df['Global']>0.01, cols_list] = df.loc[df['Global']>0.01, cols_list]/1000.0

    return df



def read_in_data(regional_measures_file, names_file, covars_file=None, names_308_style=True):
    '''
    Read in the data from the three input files:
        * regional_measures_file
        * names_file
        * covars_file

    If the names are in 308 style then drop the first 41 entries from the names
    and covars files.
    '''
    import pandas as pd
    
    # Load the input files
    df = pd.read_csv(regional_measures_file)
    with open(names_file) as f:
        names = [ line.strip() for line in f ]

    if covars_file:
        with open(covars_file) as f:
            covars_list = [ line.strip() for line in f ]
    else:
        covars_list = []

    # If you have your names in names_308_style you need to strip the
    # first 41 items
    if names_308_style:
        names = names[41:]

    # You may also have to strip the words "thickness" from the
    # end of the names in the data frame
    if names_308_style:
        df.columns = [ col.rsplit('_thickness', 1)[0] for col in df.columns ]

    return df, names, covars_list


def residuals(x, y):
    '''
    A useful little function that correlates
    x and y together to give their residual
    values. These can then be used to calculate
    partial correlation values
    '''
    import numpy as np

    if len(x.shape) == 1:
        x = x[np.newaxis, :]
    A = np.vstack([x, np.ones(x.shape[-1])]).T
    B = np.linalg.lstsq(A, y)[0]
    m = B[:-1]
    c = B[-1]
    pre = np.sum(m * x.T, axis=1) + c
    res = y - pre
    return res