script for performing full analysis and output of clustering results.

solventmapcreator/clusteringanalysis/clusteranalysis.py

+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+Script for performing the linkage calculations and outputting them to file
+and plots.
+
+@author: mark
+"""
+
+import logging
+
+import solventmapcreator.clusteringanalysis.clustercalculation as clustercalculation
+import solventmapcreator.clusteringanalysis.clusterstatistics as clusterstatistics
+import solventmapcreator.clusteringanalysis.dendrogramplotting as dendrogramplotting
+import solventmapcreator.io.linkagewriter as linkagewriter
+
+logging.basicConfig()
+LOGGER = logging.getLogger(__name__)
+LOGGER.setLevel(logging.WARN)
+
+def create_dend_and_stats_all_methods(condensed_matrix, labels, filename_stem, **kwargs):
+    """This generates the linkage matrices for each method, and returns the
+    dendrogram and linkage file and statistics for each method.
+    """
+    linkage_matrix_dict = clustering_by_all_methods(condensed_distance_matrix)
+    output_dict= {}
+    for method_label, linkage_matrix in linkage_matrix_dict.keys():
+        method_filename_stem= filename_stem + '_' + method_label
+        output_dict[method_label] = create_dendrogram_and_statistics(linkage_matrix,
+                                                                     condensed_matrix,
+                                                                     labels,
+                                                                     method_filename_stem,
+                                                                     **kwargs)
+    return output_dict
+
+def create_dendrogram_and_statistics_for_method(method, condensed_matrix, labels, filename_stem, **kwargs):
+    """This calculates the linkage matrix for the given method, then outputs the 
+    """
+    linkage_matrix = clustering_given_method(condensed_matrix, method)
+    return create_dendrogram_and_statistics(linkage_matrix, condensed_matrix,
+                                            labels, filename_stem, **kwargs)
+
+def create_dendrogram_and_statistics(linkage_matrix, condensed_matrix, labels, filename_stem, **kwargs):
+    """This generates the dendrogram and linkage files and also calculates the statistics.
+    """
+    linkage_out, dendrogram_out = create_and_write_dendrogram_and_link_file(linkage_matrix,
+                                                                            labels,
+                                                                            filename_stem,
+                                                                            **kwargs)
+    coph_filename = generate_cophenetic_filename(filename_stem)
+    coph_out = generate_cophenetic_and_inconsistency_file(linkage_matrix,
+                                                          condensed_matrix,
+                                                          coph_filename, **kwargs)
+    return linkage_out, dendrogram_out, coph_out
+
+def create_and_write_dendrogram_and_link_file(linkage_matrix, labels, filename_stem, **kwargs):
+    """This generates the dendrogram and the linkage file.
+    """
+    linkage_filename = generate_linkage_filename(filename_stem)
+    dendrogram_out = create_and_write_dendogram(linkage_matrix, labels, filename_stem, **kwargs)
+    linkage_out = write_linkage_matrix(linkage_matrix, linkage_filename)
+    return linkage_out, dendrogram_out
+
+def generate_cophenetic_and_inconsistency_file(linkage_matrix,
+                                               condensed_distance_matrix,
+                                               filename, **kwargs):
+    """This calculates the cophenetic distances and coefficient for the given matrix
+    and the inconsistency statistics and outputs to file.
+    """
+    coph_incon_dict = calculate_cophenetic_and_inconsistency(linkage_matrix,
+                                                             condensed_distance_matrix,
+                                                             **kwargs)
+    return write_cophenetic_and_inconsistency(coph_incon_dict["cophenetic_coeff"],
+                                              coph_incon_dict["cophenetic_distances"],
+                                              coph_incon_dict["inconsistency_matrix"],
+                                              filename)
+
+def generate_cophenetic_filename(filename_stem):
+    """This generates the cophenetic and inconsistency filenamefrom the given
+    stem.
+    """
+    return filename_stem + '_coph_inconst.csv
+
+def generate_linkage_filename(filename_stem):
+    """This generates the filename for the linkage file.
+    """
+    return filename_stem + '.csv'
+
+def create_and_write_dendogram(linkage_matrix, labels, filename_stem, **kwargs):
+    """This creates a dendogram plot and writes to file.
+    """
+    return dendrogramplotting.create_and_write_dendogram(linkage_matrix, labels,
+                                                         filename_stem, **kwargs)
+
+def write_linkage_matrix(linkage_matrix, filename):
+    """This writes the linkage matrix to file.
+    """
+    return linkagewriter.write_linkage_matrix(linkage_matrix, filename)
+
+def write_cophenetic_and_inconsistency(cophenetic_coefficient,
+                                       cophenetic_distances,
+                                       inconsistency_matrix, filename):
+    """This writes out the cophenetic and inconsistency information.
+    """
+    return linkagewriter.write_cophenetic_and_inconsistency(cophenetic_coefficient,
+                                                            cophenetic_distances,
+                                                            inconsistency_matrix,
+                                                            filename)
+
+def calculate_cophenetic_and_inconsistency(linkage_matrix, condensed_distance_matrix,
+                                           **kwargs):
+    """This calculates the cophenetic distances and coefficient for the given matrix
+    and the inconsistency statistics.
+    """
+    cophenetic_coeff, cophenetic_distances = calculate_cophenetic_distance(linkage_matrix,
+                                                                           condensed_distance_matrix)
+    inconsistency_matrix = calculate_inconsistency(linkage_matrix, **kwargs)
+    return {"cophenetic_coeff":cophenetic_coeff,
+            "cophenetic_distances":cophenetic_distances,
+            "inconsistency_matrix":inconsistency_matrix}
+
+def calculate_inconsistency(linkage_matrix, **kwargs):
+    """This calculates the inconsistency statistics on the linkage matrix
+    """
+    return clusterstatistics.calculate_inconsistency(linkage_matrix, kwargs.get("d", 2))
+
+def calculate_cophenetic_distance(linkage_matrix, condensed_distance_matrix):
+    """This calculates the cophenetic distances and coefficient for the given matrix.
+    """
+    return clusterstatistics.calculate_cophenetic_distance(linkage_matrix, condensed_distance_matrix)
+
+def clustering_given_method(condensed_distance_matrix, method):
+    """This calculates the linkage matrix using the given method only.
+    """
+    return clustercalculation.calculate_clustering(condensed_distance_matrix, method)
+
+def clustering_by_all_methods(condensed_distance_matrix):
+    """This calculates the clustering for each method.
+    """
+    return clustercalculation.clustering_by_all_methods(condensed_distance_matrix)