Merge pull request #81 from simonvh/seed

set random seed for clustering, closes #80

fluff/commands/heatmap.py

@@ -50,7 +50,8 @@ def heatmap(args):
     dynam = args.graphdynamics
     fontsize = args.textfontsize
     colorbar = args.colorbar
-
+    seed = args.seed
+
     # Check for mutually exclusive parameters
     if dynam:
         if merge_mirrored:
@@ -147,7 +148,8 @@ def load_data(featurefile, amount_bins, extend_dyn_up, extend_dyn_down, rmdup, r
             clus, 
             cluster_type=cluster_type, 
             numclusters = args.numclusters, 
-            dist=METRIC)
+            dist=METRIC,
+            random_state=seed)
 
     if cluster_type == "k":
        if not dynam and merge_mirrored:

fluff/config.py

@@ -1,5 +1,5 @@
 ### CONSTANTS ###
-FL_VERSION = "3.0.2"
+FL_VERSION = "3.0.3"
 
 ######## HEATMAP #############
 DEFAULT_BINSIZE = 25 

fluff/parse.py

@@ -82,7 +82,11 @@ def parse_cmds():
                          help="pick specific data files to use for clustering",
                          default=None,
                          type=str)
-
+    clu_grp.add_argument("-S",
+                         dest="seed",
+                         help="random seed (int) to use for K-means clustering",
+                         default=None,
+                         type=int)
     dap_grp = p.add_argument_group(title='Data processing')
     dap_grp.add_argument("-r",
                          dest="rpkm",
@@ -362,4 +366,4 @@ def main():
         main()
     except KeyboardInterrupt:
         sys.stderr.write("User interrupted me! ;-) Bye!\n")
-        sys.exit(0)
+        sys.exit(0)

fluff/util.py

@@ -187,7 +187,7 @@ def mirror_clusters(data, labels, cutoff=0.01):
     return (None, None)
 
 
-def cluster_profile(cluster_data, cluster_type="k", numclusters=3, dist="euclidean"):
+def cluster_profile(cluster_data, cluster_type="k", numclusters=3, dist="euclidean", random_state=None):
     """Cluster profiles for heatmap
 
     Takes a matrix and clusters either with kmeans or hierarchical clustering.
@@ -227,7 +227,7 @@ def cluster_profile(cluster_data, cluster_type="k", numclusters=3, dist="euclide
         print("K-means clustering")
         ## K-means clustering
 
-        k = KMeans(n_clusters=numclusters)
+        k = KMeans(n_clusters=numclusters, random_state=random_state)
         labels = k.fit(cluster_data).labels_
         ind = labels.argsort()