Analyzing the correlation matrices

In [ ]:

    

%pylab inline
from __future__ import print_function, division

import os
import os.path as osp
import matplotlib.pyplot as plt

from warnings import warn
import datetime, time
import glob as gb
from six import string_types
import argparse
import json
import time


import numpy as np
import scipy.linalg as lin
import scipy.stats as sst

import nibabel as nib
import nipy
from nilearn._utils import concat_niimgs


    

In [ ]:

    

import importlib
from smpce_data_to_corr import get_params
import utils._utils as ucr
import utils.setup_filenames as suf
import correlation2results as c2r
#import tests.test_smpce_data_to_corr as tts

ucr = reload(ucr)
suf = reload(suf)
c2r = reload(c2r)


    

In [ ]:

    

param_dir = osp.abspath('.')
assert param_dir == '/home/jb/code/simpace/simpace'
# or assert on neuro
params = get_params(param_dir)
basedir = '/home/jb/data/simpace/data/rename_files'


    

In [ ]:

    

djdata = params['data']
djlayo = params['layout']

nb_sess = params['data']['nb_sess']
nb_sub = params['data']['nb_sub']
nb_run = params['data']['nb_run']
print(nb_sub, nb_sess, nb_run)


    

In [ ]:

    

def _get_common_labels(conds, idx0=0):
    
    cond0 = conds.keys()[idx0]
    print(cond0)
    nb_sess = len(conds[cond0])
    
    lsets = []
    for sess in range(nb_sess):
        lsets.append( set((np.load(conds[cond0][sess]))['labels_sig']) )

    return set.intersection(*lsets)


    

In [ ]:

    

conds = c2r._get_signals_filenames(basedir, params)
aaa = _get_common_labels(conds)
bbb = _get_common_labels(conds, idx0=3)
ccc = _get_common_labels(conds, idx0=2)
assert aaa == bbb
assert aaa == ccc


    

In [ ]:

    

print("\n".join(conds['none'][8:]))
conds.keys()


    

In [ ]:

    

conds = c2r._get_signals_filenames(basedir, params)
print(conds.keys())
common_labels = c2r._get_common_labels(conds)
assert common_labels == _get_common_labels(conds, idx0=3)


    

In [ ]:

    

conds_arr, stored_param = c2r.compute_corr_mtx(conds, common_labels)


    

In [ ]:

    

c2r.save_results(basedir, "results", "all_corrections", params)


    

In [ ]:

    

tmp = np.load(conds['med'][7])
#tmp.keys()


    

In [ ]:

    

tmp['labels_sig'].shape
tmp['arr_sig_f'].shape


    

In [ ]:

    




    

In [ ]:

    

print(conds_arr.keys())
print(conds_arr['high'].shape)
for cond in ordered_conds(): # [none_c, low_c, med_c, high_c]:
    assert np.all(conds_arr[cond] <= 1.) and np.all(conds_arr[cond] >= -1.)


    

In [ ]:

    

f, axes = plt.subplots(1, 4)
arr = [conds_arr[c] for c in ordered_conds()]

for idx, ax in enumerate(axes): 
    ax.imshow(arr[idx].mean(axis=0), aspect='equal', interpolation='nearest', 
              vmin=-.5, vmax=1.)


    

In [ ]:

    

f, axes = plt.subplots(1, 4)
arr = [conds_arr[c] for c in ordered_conds()]

for idx, ax in enumerate(axes): 
    ax.imshow(arr[idx].mean(axis=0) - arr[0].mean(axis=0), 
              aspect='equal', interpolation='nearest',
              vmin=-.5, vmax=.5)


    

In [ ]:

    

a0 = conds_arr['none'].mean(axis=0)
for idx, a in enumerate(arr):
    print((a.mean(axis=0) - a0).min(), (a.mean(axis=0) - a0).max())


    

In [ ]:

    

f, axes = plt.subplots(1, 4)
arr = [conds_arr[c] for c in ordered_conds()]

for idx, ax in enumerate(axes): 
    ax.imshow(arr[idx].std(axis=0), 
              aspect='equal', interpolation='nearest') #, vmin=0., vmax=.5)


    

In [ ]:

    

for idx, a in enumerate(arr):
    print((a.std(axis=0)).mean())