Merge branch 'master' of https://github.com/qiime/emperor into pep8-util

biocore · Mar 23, 2015 · d0a2d3a · d0a2d3a
2 parents 3e7fecc + b74c666
commit d0a2d3a
Show file tree

Hide file tree

Showing 2 changed files with 133 additions and 101 deletions.
diff --git a/emperor/sort.py b/emperor/sort.py
@@ -1,41 +1,37 @@
-#!/usr/bin/env python
-# File created on 20 Apr 2013
 from __future__ import division
-
-__author__ = "Yoshiki Vazquez Baeza"
-__copyright__ = "Copyright 2013, The Emperor Project"
-__credits__ = ["Yoshiki Vazquez Baeza"]
-__license__ = "BSD"
-__version__ = "0.9.51-dev"
-__maintainer__ = "Yoshiki Vazquez Baeza"
-__email__ = "yoshiki89@gmail.com"
-__status__ = "Development"
-
 from numpy import zeros
 from re import compile, search
 
+
 def sort_taxa_table_by_pcoa_coords(coords_header, otu_table, otu_header):
     """Sort and match the samples in the otu table and in the coordinates data
 
-    Inputs:
-    coords_header: sample ids that are present in principal coordinates data
-    otu_table: numpy array with the data for an otu table
-    otu_header: sample ids present in the otu table
-
-    Ouputs:
-    sorted_otu_headers: sample ids that were present in the coords_header list,
-    the order in this table matches the order of the coordinates data
-    sorted_otu_table: otu table data with columns belonging to the sample ids in
-    the sorted_otu_headers list
-
     This function will sort the columns of an otu table as suggested by the
     sample ids in the coords_header
+
+    Parameters
+    ----------
+    coords_header: list of str
+        sample ids that are present in principal coordinates data
+    otu_table: numpy array
+        numpy array with the data for an otu table
+    otu_header: list of str
+        sample ids present in the otu table
+
+    Returns
+    -------
+    sorted_otu_headers: list of str
+        sample ids that were present in the coords_header list, the order in
+        this table matches the order of the coordinates data
+    sorted_otu_table: numpy array
+        otu table data with columns belonging to the sample ids in the
+        sorted_otu_headers list
     """
 
     sorted_otu_headers = []
 
     # the size of the otu table can be pre-allocated for better memory usage
-    matching_headers = len(set(coords_header)&set(otu_header))
+    matching_headers = len(set(coords_header) & set(otu_header))
     sorted_otu_table = zeros([otu_table.shape[0], matching_headers])
 
     # iterate through the available sample ids in the coordinates file and work
@@ -44,31 +40,36 @@ def sort_taxa_table_by_pcoa_coords(coords_header, otu_table, otu_header):
     for i, element in enumerate(coords_header):
         if element in otu_header:
             current_index = otu_header.index(element)
-            sorted_otu_table[:,i] = otu_table[:,current_index]
+            sorted_otu_table[:, i] = otu_table[:, current_index]
             sorted_otu_headers.append(element)
 
     return sorted_otu_headers, sorted_otu_table
 
+
 def sort_comparison_filenames(coord_fps):
     """Pass in a list of file names and sort them using the suffix
 
-    Input:
-    coord_fps: list of filenames with the format something_something_qX.txt
-    where X is the index of the file.
-
-    Output:
-    Returns a sorted version of the list that was passed in where the strings
-    are sorted according to the suffix they have, if the string doesn't have
-    a suffix it will be added to the beginning of the list.
+    Parameters
+    ----------
+    coord_fps: list of str
+        The filenames with the format something_something_qX.txt where X is
+        the index of the file.
+
+    Returns
+    -------
+    list of str
+        A sorted version of the list that was passed in where the strings are
+        sorted according to the suffix they have, if the string doesn't have a
+        suffix it will be added to the beginning of the list.
     """
 
     if coord_fps == []:
         return []
 
     def _get_suffix(fp):
         """Gets the number in the suffix for a string using a regex"""
-        # any alphanumeric set of characters proceeded by a 'q', a number, a dot
-        # & a txt extension at the end of the line. Take for example 
+        # any alphanumeric set of characters proceeded by a 'q', a number,
+        # a dot & a txt extension at the end of the line. Take for example
         # bray_curtis_q1.txt or unifrac_q11.txt
         re = compile(r'(\w+)_q([0-9]+).txt$')
         tmatch = search(re, fp)
@@ -84,4 +85,3 @@ def _get_suffix(fp):
     # the key function retrieves the suffix number for the function to sort
     # according to it's floating point representation i. e. the cast to float
     return sorted(coord_fps, key=_get_suffix)
-
diff --git a/tests/test_sort.py b/tests/test_sort.py
@@ -1,46 +1,43 @@
 #!/usr/bin/env python
-# File created on 20 Apr 2013
-from __future__ import division
-
-__author__ = "Yoshiki Vazquez Baeza"
-__copyright__ = "Copyright 2013, The Emperor Project"
-__credits__ = ["Yoshiki Vazquez Baeza"]
-__license__ = "BSD"
-__version__ = "0.9.51-dev"
-__maintainer__ = "Yoshiki Vazquez Baeza"
-__email__ = "yoshiki89@gmail.com"
-__status__ = "Development"
 
+from __future__ import division
 from unittest import TestCase, main
 
 from numpy import array
 from numpy.testing import assert_almost_equal
 
 from emperor.sort import (sort_taxa_table_by_pcoa_coords,
-    sort_comparison_filenames)
+                          sort_comparison_filenames)
+
 
 class TopLevelTests(TestCase):
     def setUp(self):
         self.otu_headers = ['PC.636', 'PC.635', 'PC.356', 'PC.481', 'PC.354',
-                'PC.593', 'PC.355', 'PC.607', 'PC.634']
-
-        self.otu_table = array([[0.02739726, 0.04697987, 0.02, 0.04697987, 0.01,
-            0.02027027, 0.01360544, 0.01342282, 0.02666667], [0.00684932,
-            0.02013423, 0.02, 0.00671141,  0., 0.00675676, 0., 0., 0.], [
-            0.14383562, 0.27516779, 0.65333333, 0.52348993, 0.38926174,
-            0.69594595, 0.28571429, 0.0738255, 0.19333333], [0., 0.02013423,
-            0.03333333, 0.01342282, 0., 0.0472973, 0., 0., 0.], [0.78767123,
-            0.45637584, 0.22, 0.39597315, 0.41610738, 0.20945946, 0.70068027,
-            0.89932886, 0.77333333], [0.,0.02013423, 0.01333333, 0.00671141,
-            0.03355705, 0.00675676, 0., 0., 0.],[0., 0., 0.01333333, 0., 0., 0.,
-            0., 0., 0.], [0.03424658, 0.16107383, 0.02666667, 0.00671141,
-            0.14765101, 0.01351351, 0., 0.01342282, 0.00666667]])
+                            'PC.593', 'PC.355', 'PC.607', 'PC.634']
+
+        self.otu_table = array(
+            [[0.02739726, 0.04697987, 0.02, 0.04697987, 0.01, 0.02027027,
+              0.01360544, 0.01342282, 0.02666667],
+             [0.00684932, 0.02013423, 0.02, 0.00671141,  0., 0.00675676, 0.,
+              0., 0.],
+             [0.14383562, 0.27516779, 0.65333333, 0.52348993, 0.38926174,
+              0.69594595, 0.28571429, 0.0738255, 0.19333333],
+             [0., 0.02013423, 0.03333333, 0.01342282, 0., 0.0472973, 0.,
+              0., 0.],
+             [0.78767123, 0.45637584, 0.22, 0.39597315, 0.41610738, 0.20945946,
+              0.70068027, 0.89932886, 0.77333333],
+             [0., 0.02013423, 0.01333333, 0.00671141, 0.03355705, 0.00675676,
+              0., 0., 0.],
+             [0., 0., 0.01333333, 0., 0., 0., 0., 0., 0.],
+             [0.03424658, 0.16107383, 0.02666667, 0.00671141, 0.14765101,
+              0.01351351, 0., 0.01342282, 0.00666667]])
 
         self.coords = COORDS
-        self.coords_header = ['PC.354','PC.356','PC.481','PC.593',
-            'PC.355','PC.607','PC.634', 'PC.636', 'PC.635']
+        self.coords_header = ['PC.354', 'PC.356', 'PC.481', 'PC.593', 'PC.355',
+                              'PC.607', 'PC.634', 'PC.636', 'PC.635']
 
-        self.coord_fps = ['output_data/emperor/bray_curtis_pc_transformed_q1.txt',
+        self.coord_fps = [
+            'output_data/emperor/bray_curtis_pc_transformed_q1.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q10.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q11.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q12.txt',
@@ -77,15 +74,18 @@ def setUp(self):
             'output_data/emperor/bray_curtis_pc_transformed_q12.txt',
             'output_data/emperor/qqq2223_curtis_qc_transformed_q13.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q14.txt',
-            'output_data/emperor/bray_curtis_pc_transformed_reference.txtoutput_data/emperor/bray_curtis_pc_transformed_q15.txt',
+            'output_data/emperor/bray_curtis_pc_transformed_reference.'
+            'txtoutput_data/emperor/bray_curtis_pc_transformed_q15.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q16.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q17.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q18.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q19.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q2.txt',
             'output_data/emperor/boom.txt',
-            'output_data/emperor/another_file with some characters and stuff .txt',
-            'output_data/emperor/some_other_file_that_foo_wants_to_compare.txt',
+            'output_data/emperor/another_file with some characters '
+            'and stuff .txt',
+            'output_data/emperor/some_other_file_that_foo_wants_to_'
+            'compare.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q23.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q24.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q25.txt',
@@ -101,28 +101,31 @@ def setUp(self):
             'output_data/emperor/bray_curtis_pc_transformed_q8.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q9.txt']
 
-
-
     def test_sort_taxa_table_by_pcoa_coords(self):
         """Make sure OTU table and coordinates are sorted equally"""
 
         # case with shuffled inputs
         o_headers, o_otu_table = sort_taxa_table_by_pcoa_coords(
             self.coords_header, self.otu_table, self.otu_headers)
 
-        self.assertEquals(o_headers, ['PC.354','PC.356','PC.481','PC.593',
-            'PC.355','PC.607','PC.634', 'PC.636', 'PC.635'])
+        self.assertEquals(o_headers, ['PC.354', 'PC.356', 'PC.481', 'PC.593',
+                                      'PC.355', 'PC.607', 'PC.634', 'PC.636',
+                                      'PC.635'])
         assert_almost_equal(o_otu_table, OTU_TABLE_A)
 
         # case with shuffled inputs and fewer samples
-        o_headers, o_otu_table = sort_taxa_table_by_pcoa_coords(['PC.354',
-            'PC.356','PC.635'], self.otu_table, self.otu_headers)
-        self.assertEquals(o_headers, ['PC.354','PC.356','PC.635'])
-        assert_almost_equal(o_otu_table, array([[ 0.01, 0.02, 0.04697987],[0.,
-            0.02, 0.02013423], [0.38926174, 0.65333333, 0.27516779],[0.,
-            0.03333333, 0.02013423],[0.41610738, 0.22, 0.45637584],[0.03355705,
-            0.01333333, 0.02013423],[0., 0.01333333, 0.],[0.14765101,
-            0.02666667, 0.16107383]]))
+        o_headers, o_otu_table = sort_taxa_table_by_pcoa_coords(
+            ['PC.354', 'PC.356', 'PC.635'], self.otu_table, self.otu_headers)
+        self.assertEquals(o_headers, ['PC.354', 'PC.356', 'PC.635'])
+        assert_almost_equal(o_otu_table, array(
+            [[0.01, 0.02, 0.04697987],
+             [0., 0.02, 0.02013423],
+             [0.38926174, 0.65333333, 0.27516779],
+             [0., 0.03333333, 0.02013423],
+             [0.41610738, 0.22, 0.45637584],
+             [0.03355705, 0.01333333, 0.02013423],
+             [0., 0.01333333, 0.],
+             [0.14765101, 0.02666667, 0.16107383]]))
 
     def test_sort_comparison_filenames_regular(self):
         """Check filenames are sorted correctly"""
@@ -163,10 +166,13 @@ def test_sort_comparison_filenames_regular(self):
         # if files with garbage are passed in, the sorting should be still
         # consistent,putting the "garbaged" filenames at the beginning
         out_sorted = sort_comparison_filenames(self.coord_fps_garbage)
-        self.assertEquals(out_sorted, ['output_data/emperor/aaaaaaa.txt',
+        self.assertEquals(out_sorted, [
+            'output_data/emperor/aaaaaaa.txt',
             'output_data/emperor/boom.txt',
-            'output_data/emperor/another_file with some characters and stuff .txt',
-            'output_data/emperor/some_other_file_that_foo_wants_to_compare.txt',
+            'output_data/emperor/another_file with some characters and '
+            'stuff .txt',
+            'output_data/emperor/some_other_file_that_foo_wants_to_'
+            'compare.txt',
             'output_data/emperor/bray_qurtis_pc_transformed_q1.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q2.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q3.txt',
@@ -180,7 +186,8 @@ def test_sort_comparison_filenames_regular(self):
             'output_data/emperor/bray_curtis_pc_transformed_q12.txt',
             'output_data/emperor/qqq2223_curtis_qc_transformed_q13.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q14.txt',
-            'output_data/emperor/bray_curtis_pc_transformed_reference.txtoutput_data/emperor/bray_curtis_pc_transformed_q15.txt',
+            'output_data/emperor/bray_curtis_pc_transformed_reference.'
+            'txtoutput_data/emperor/bray_curtis_pc_transformed_q15.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q16.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q17.txt',
             'output_data/emperor/bray_curtis_pc_transformed_q18.txt',
@@ -211,24 +218,49 @@ def test_sort_comparison_filenames_regular(self):
         self.assertEquals(sort_comparison_filenames([]), [])
 
 
-COORDS = array([[0.280399117569, -0.0060128286014, 0.0234854344148, -0.0468109474823, -0.146624450094, 0.00566979124596, -0.0354299634191, -0.255785794275, -4.84141986706e-09],
-[0.228820399536, -0.130142097093, -0.287149447883, 0.0864498846421, 0.0442951919304, 0.20604260722, 0.0310003571386, 0.0719920436501, -4.84141986706e-09],
-[0.0422628480532, -0.0139681511889, 0.0635314615517, -0.346120552134, -0.127813807608, 0.0139350721063, 0.0300206887328, 0.140147849223, -4.84141986706e-09],
-[0.232872767451, 0.139788385269, 0.322871079774, 0.18334700682, 0.0204661596818, 0.0540589147147, -0.0366250872041, 0.0998235721267, -4.84141986706e-09],
-[0.170517581885, -0.194113268955, -0.0308965283066, 0.0198086158783, 0.155100062794, -0.279923941712, 0.0576092515759, 0.0242481862127, -4.84141986706e-09],
-[-0.0913299284215, 0.424147148265, -0.135627421345, -0.057519480907, 0.151363490722, -0.0253935675552, 0.0517306152066, -0.038738217609, -4.84141986706e-09],
-[-0.349339228244, -0.120787589539, 0.115274502117, 0.0694953933826, -0.0253722182853, 0.067853201946, 0.244447634756, -0.0598827706386, -4.84141986706e-09],
-[-0.276542163845, -0.144964375408, 0.0666467344429, -0.0677109454288, 0.176070269506, 0.072969390136, -0.229889463523, -0.0465989416581, -4.84141986706e-09],
-[-0.237661393984, 0.0460527772512, -0.138135814766, 0.159061025229, -0.247484698646, -0.115211468101, -0.112864033263, 0.0647940729676, -4.84141986706e-09]])
-
-OTU_TABLE_A = array([[ 0.01, 0.02, 0.04697987, 0.02027027, 0.01360544, 0.01342282, 0.02666667, 0.02739726, 0.04697987],
-[ 0., 0.02, 0.00671141, 0.00675676, 0., 0., 0., 0.00684932, 0.02013423],
-[ 0.38926174, 0.65333333, 0.52348993, 0.69594595, 0.28571429, 0.0738255, 0.19333333, 0.14383562, 0.27516779],
-[ 0., 0.03333333, 0.01342282, 0.0472973, 0., 0., 0., 0., 0.02013423],
-[ 0.41610738, 0.22, 0.39597315, 0.20945946, 0.70068027, 0.89932886, 0.77333333, 0.78767123, 0.45637584],
-[ 0.03355705, 0.01333333, 0.00671141, 0.00675676, 0., 0., 0., 0., 0.02013423],
-[ 0., 0.01333333, 0., 0., 0., 0., 0., 0., 0.],
-[ 0.14765101, 0.02666667, 0.00671141, 0.01351351, 0., 0.01342282, 0.00666667, 0.03424658, 0.16107383]])
+COORDS = array(
+    [[0.280399117569, -0.0060128286014, 0.0234854344148, -0.0468109474823,
+      -0.146624450094, 0.00566979124596, -0.0354299634191, -0.255785794275,
+      -4.84141986706e-09],
+     [0.228820399536, -0.130142097093, -0.287149447883, 0.0864498846421,
+      0.0442951919304, 0.20604260722, 0.0310003571386, 0.0719920436501,
+      -4.84141986706e-09],
+     [0.0422628480532, -0.0139681511889, 0.0635314615517, -0.346120552134,
+      -0.127813807608, 0.0139350721063, 0.0300206887328, 0.140147849223,
+      -4.84141986706e-09],
+     [0.232872767451, 0.139788385269, 0.322871079774, 0.18334700682,
+      0.0204661596818, 0.0540589147147, -0.0366250872041, 0.0998235721267,
+      -4.84141986706e-09],
+     [0.170517581885, -0.194113268955, -0.0308965283066, 0.0198086158783,
+      0.155100062794, -0.279923941712, 0.0576092515759, 0.0242481862127,
+      -4.84141986706e-09],
+     [-0.0913299284215, 0.424147148265, -0.135627421345, -0.057519480907,
+      0.151363490722, -0.0253935675552, 0.0517306152066, -0.038738217609,
+      -4.84141986706e-09],
+     [-0.349339228244, -0.120787589539, 0.115274502117, 0.0694953933826,
+      -0.0253722182853, 0.067853201946, 0.244447634756, -0.0598827706386,
+      -4.84141986706e-09],
+     [-0.276542163845, -0.144964375408, 0.0666467344429, -0.0677109454288,
+      0.176070269506, 0.072969390136, -0.229889463523, -0.0465989416581,
+      -4.84141986706e-09],
+     [-0.237661393984, 0.0460527772512, -0.138135814766, 0.159061025229,
+      -0.247484698646, -0.115211468101, -0.112864033263, 0.0647940729676,
+      -4.84141986706e-09]])
+
+OTU_TABLE_A = array(
+    [[0.01, 0.02, 0.04697987, 0.02027027, 0.01360544, 0.01342282, 0.02666667,
+      0.02739726, 0.04697987],
+     [0., 0.02, 0.00671141, 0.00675676, 0., 0., 0., 0.00684932, 0.02013423],
+     [0.38926174, 0.65333333, 0.52348993, 0.69594595, 0.28571429, 0.0738255,
+      0.19333333, 0.14383562, 0.27516779],
+     [0., 0.03333333, 0.01342282, 0.0472973, 0., 0., 0., 0., 0.02013423],
+     [0.41610738, 0.22, 0.39597315, 0.20945946, 0.70068027, 0.89932886,
+      0.77333333, 0.78767123, 0.45637584],
+     [0.03355705, 0.01333333, 0.00671141, 0.00675676, 0., 0., 0., 0.,
+      0.02013423],
+     [0., 0.01333333, 0., 0., 0., 0., 0., 0., 0.],
+     [0.14765101, 0.02666667, 0.00671141, 0.01351351, 0., 0.01342282,
+      0.00666667, 0.03424658, 0.16107383]])
 
 
 if __name__ == "__main__":