Merge pull request #356 from cosanlab/srm

added isfc

Author: ljchang

nltools/data/adjacency.py

@@ -26,7 +26,6 @@
                            two_sample_permutation,
                            summarize_bootstrap,
                            matrix_permutation,
-                           jackknife_permutation,
                            fisher_r_to_z,
                            _calc_pvalue,
                            _bootstrap_isc)
@@ -390,8 +389,8 @@ def plot(self, limit=3, axes=None, *args, **kwargs):
         ''' Create Heatmap of Adjacency Matrix
             
             Can pass in any sns.heatmap argument
-            
-            Args: 
+
+            Args:
                 limit: (int) number of heatmaps to plot if object contains multiple adjacencies (default: 3)
                 axes: matplotlib axis handle
         '''
@@ -576,7 +575,7 @@ def similarity(self, data, plot=False, perm_type='2d', n_permute=5000,
         Default is to use spearman correlation and permutation test.
         Args:
             data: Adjacency data, or 1-d array same size as self.data
-            perm_type: (str) '1d','2d', 'jackknife', or None
+            perm_type: (str) '1d','2d', or None
             metric: (str) 'spearman','pearson','kendall'
             ignore_diagonal: (bool) only applies to 'directed' Adjacency types using perm_type=None or perm_type='1d'
         '''
@@ -593,10 +592,8 @@ def similarity(self, data, plot=False, perm_type='2d', n_permute=5000,
             similarity_func = correlation_permutation
         elif perm_type == '2d':
             similarity_func = matrix_permutation
-        elif perm_type == 'jackknife':
-            similarity_func = jackknife_permutation
         else:
-            raise ValueError("perm_type must be ['1d','2d', 'jackknife', or None']")
+            raise ValueError("perm_type must be ['1d','2d', or None']")
 
         def _convert_data_similarity(data, perm_type=None, ignore_diagonal=ignore_diagonal):
             '''Helper function to convert data correctly'''
@@ -606,13 +603,13 @@ def _convert_data_similarity(data, perm_type=None, ignore_diagonal=ignore_diagon
                     data = d[~np.eye(d.shape[0]).astype(bool)]
                 else:
                     data = data.data
-            elif perm_type in ['2d', 'jackknife']:
+            elif perm_type == '2d':
                 if not data.issymmetric:
                     raise TypeError(f"data must be symmetric to do {perm_type} permutation")
                 else:
                     data = data.squareform()
             else:
-                raise ValueError("perm_type must be ['1d','2d', 'jackknife', or None']")
+                raise ValueError("perm_type must be ['1d','2d', or None']")
             return data
 
         if self.is_single_matrix:

nltools/data/brain_data.py

@@ -18,6 +18,7 @@
 from scipy.stats import ttest_1samp, pearsonr
 from scipy.stats import t as t_dist
 from scipy.signal import detrend
+from scipy.interpolate import pchip
 import os
 import shutil
 import nibabel as nib
@@ -1887,6 +1888,41 @@ def find_spikes(self, global_spike_cutoff=3, diff_spike_cutoff=3):
                             global_spike_cutoff=global_spike_cutoff,
                             diff_spike_cutoff=diff_spike_cutoff)
 
+    def temporal_resample(self, sampling_freq=None, target=None, target_type='hz'):
+        ''' Resample Brain_Data timeseries to a new target frequency or number of samples
+        using Piecewise Cubic Hermite Interpolating Polynomial (PCHIP) interpolation.
+        This function can up- or down-sample data.
+        
+        Note: this function can use quite a bit of RAM.
+
+            Args:
+                sampling_freq:  (float) sampling frequency of data in hertz
+                target: (float) upsampling target
+                target_type: (str) type of target can be [samples,seconds,hz]
+                
+            Returns:
+                upsampled Brain_Data instance
+        '''
+        
+        out = self.copy()
+
+        if target_type == 'samples':
+            n_samples = target
+        elif target_type == 'seconds':
+            n_samples = target*sampling_freq
+        elif target_type == 'hz':
+            n_samples = float(sampling_freq)/float(target)
+        else:
+            raise ValueError('Make sure target_type is "samples", "seconds", or "hz".')
+        
+        orig_spacing = np.arange(0, self.shape()[0], 1)
+        new_spacing = np.arange(0, self.shape()[0], n_samples)
+        
+        out.data = np.zeros([len(new_spacing), self.shape()[1]])
+        for i in range(self.shape()[1]):
+            interpolate = pchip(orig_spacing, self.data[:, i])
+            out.data[:, i] = interpolate(new_spacing)
+        return out
 
 class Groupby(object):
     def __init__(self, data, mask):

nltools/mask.py

@@ -43,6 +43,7 @@ def create_sphere(coordinates, radius=5, mask=None):
             else:
                 raise ValueError("mask is not a nibabel instance or a valid "
                                  "file name")
+
     else:
         mask = nib.load(resolve_mni_path(MNI_Template)['mask'])
 
@@ -182,33 +183,19 @@ def roi_to_brain(data, mask_x):
     Args:
         data: Pandas series, dataframe, list, np.array of ROI by observation
         mask_x: an expanded binary mask
+
     Returns:
         out: (Brain_Data) Brain_Data instance where each ROI is now populated
              with a value
     '''
     from nltools.data import Brain_Data
 
-    def series_to_brain(data, mask_x):
-        '''Converts a pandas series of ROIs to a Brain_Data instance. Index must correspond to ROI index'''
-
-        if not isinstance(data, pd.Series):
-            raise ValueError('Data must be a pandas series')
-        if len(mask_x) != len(data):
-            raise ValueError('Data must have the same number of rows as mask has ROIs.')
-        return Brain_Data([mask_x[x]*data[x] for x in data.keys()]).sum()
-
     if not isinstance(data, (pd.Series, pd.DataFrame)):
         if isinstance(data, list):
-            if len(data) != len(mask_x):
-                raise ValueError('Data must have the same number of rows as mask has ROIs.')
-            else:
-                data = pd.Series(data)
+            data = pd.Series(data)
         elif isinstance(data, np.ndarray):
             if len(data.shape) == 1:
-                if len(data) != len(mask_x):
-                    raise ValueError('Data must have the same number of rows as mask has ROIs.')
-                else:
-                    data = pd.Series(data)
+                data = pd.Series(data)
             elif len(data.shape) == 2:
                 data = pd.DataFrame(data)
                 if data.shape[0] != len(mask_x):
@@ -220,15 +207,17 @@ def series_to_brain(data, mask_x):
                 raise NotImplementedError
 
         else:
-            raise NotImplementedError
-
+            raise ValueError("Data must be a pandas series or data frame.")
 
     if len(mask_x) != data.shape[0]:
         raise ValueError('Data must have the same number of rows as mask has ROIs.')
 
     if isinstance(data, pd.Series):
-        return series_to_brain(data, mask_x)
-    elif isinstance(data, pd.DataFrame):
-        return Brain_Data([series_to_brain(data[x], mask_x) for x in data.keys()])
+        return Brain_Data([mask_x[x]*data[x] for x in data.keys()]).sum()
     else:
-        raise ValueError("Data must be a pandas series or data frame.")
+        out = mask_x.copy()
+        out.data = np.ones((data.shape[1], out.data.shape[1]))
+        for roi in range(len(mask_x)):
+            roi_data = np.reshape(data.iloc[roi,:].values, (-1,1))
+            out.data[:, mask_x[roi].data==1] = np.repeat(roi_data.T, np.sum(mask_x[roi].data==1), axis=0).T
+        return out