sorting can be enabled and disabled

Author: K3rn1n4tor

clustering_api.py

@@ -100,8 +100,8 @@ def loadAttribute(jsonData, attr):
 
 ########################################################################################################################
 
-@app.route('/distances/<metric>/<datasetID>', methods=['POST'])
-def getDistances(metric, datasetID):
+@app.route('/distances/<metric>/<datasetID>/<sorted>', methods=['POST'])
+def getDistances(metric, datasetID, sorted):
     """
     Compute the distances of the current stratification values to its centroid.
     :param metric:
@@ -118,7 +118,7 @@ def getDistances(metric, datasetID):
     else:
         return ''
 
-    response = getClusterDistances(data, labels, metric, externLabels)
+    response = getClusterDistances(data, labels, metric, externLabels, sorted)
     return flask.jsonify(response)
 
 ########################################################################################################################

clustering_service.py

@@ -118,7 +118,7 @@ def runFuzzy(data, numClusters, m, threshold):
 
 ########################################################################################################################
 
-def getClusterDistances(data, labels, metric, externLabels = None):
+def getClusterDistances(data, labels, metric, externLabels = None, sorted = True):
     """
     Compute the cluster distances in a given data among certain rows (labels)
     :param data: genomic data
@@ -128,7 +128,7 @@ def getClusterDistances(data, labels, metric, externLabels = None):
     :return: labels and distances values sorted in ascending order
     """
     from clustering_util import computeClusterInternDistances, computeClusterExternDistances
-    distLabels, distValues = computeClusterInternDistances(data, labels, True, metric)
+    distLabels, distValues = computeClusterInternDistances(data, labels, sorted, metric)
 
     if externLabels is not None:
         externDists = computeClusterExternDistances(data, distLabels, externLabels, metric)

clustering_util.py

@@ -244,7 +244,7 @@ def similarityMeasurement(matrix, vector, method='euclidean'):
     if method == 'sqeuclidean':
         return euclideanDistance(matrix, vector, True)
 
-    spatialMethods = ['cityblock', 'chebyshev', 'canberra', 'correlation', 'hamming', 'mahalanobis', 'correlation']
+    spatialMethods = ['cityblock', 'chebyshev', 'canberra', 'correlation', 'hamming', 'mahalanobis',]
 
     if method in spatialMethods:
         return cdist(matrix, np.atleast_2d(vector), method).flatten()
@@ -376,7 +376,7 @@ def similarityMeasurementMatrix(matrix, method):
         return euclideanDistanceMatrix(matrix, True)
         # return squareform(pdist(matrix, method))
 
-    spatialMethods = ['cityblock', 'chebyshev', 'canberra', 'correlation', 'hamming', 'mahalanobis', 'correlation']
+    spatialMethods = ['cityblock', 'chebyshev', 'canberra', 'correlation', 'hamming', 'mahalanobis']
 
     if method in spatialMethods:
         return squareform(pdist(matrix, method))
@@ -411,12 +411,19 @@ def computeClusterInternDistances(matrix, labels, sorted=True, metric='euclidean
     # compute distances to centroid
     dists = similarityMeasurement(subMatrix, centroid, metric)
 
-    if sorted:
+    if sorted == 'true':
         # sort values
         indices = range(len(dists))
         indices.sort(key=dists.__getitem__)
         dists.sort()
 
+        # reverse order if correlation coefficient is used
+        # (1 means perfect correlation while -1 denotes opposite correlation)
+        corrMetrics = ['pearson', 'spearman', 'kendall']
+        if metric in corrMetrics:
+            indices.reverse()
+            dists = dists[::-1]
+
         # write back to our arrays
         distLabels = clusterLabels[indices].tolist()
         distValues = dists.tolist()