Merge pull request #37 from Avsecz/master

Added `min_metacluster_size_frac`

modisco/tfmodisco_workflow/workflow.py

@@ -143,6 +143,7 @@ def __init__(self,
                  histogram_bins=100, percentiles_in_bandwidth=10, 
                  overlap_portion=0.5,
                  min_metacluster_size=100,
+                 min_metacluster_size_frac=0.01,
                  weak_threshold_for_counting_sign=0.8,
                  max_seqlets_per_metacluster=20000,
                  target_seqlet_fdr=0.2,
@@ -163,6 +164,7 @@ def __init__(self,
         self.percentiles_in_bandwidth = percentiles_in_bandwidth
         self.overlap_portion = overlap_portion
         self.min_metacluster_size = min_metacluster_size
+        self.min_metacluster_size_frac = min_metacluster_size_frac
         self.target_seqlet_fdr = target_seqlet_fdr
         self.weak_threshold_for_counting_sign =\
             weak_threshold_for_counting_sign
@@ -234,6 +236,14 @@ def __call__(self, task_names, contrib_scores,
             print("WARNING: you found relatively few seqlets."
                   +" Consider dropping target_seqlet_fdr") 
 
+
+        if int(self.min_metacluster_size_frac * len(seqlets)) > self.min_metacluster_size:
+            print("min_metacluster_size_frac * len(seqlets) = {0} is more than min_metacluster_size={1}.".\
+                  format(int(self.min_metacluster_size_frac * len(seqlets)), self.min_metacluster_size))
+            print("Using it as a new min_metacluster_size")
+            self.min_metacluster_size = int(self.min_metacluster_size_frac * len(seqlets))
+
+
         if (self.weak_threshold_for_counting_sign is None):
             weak_threshold_for_counting_sign = weakest_transformed_thresh
         else:

setup.py

@@ -6,7 +6,7 @@
           description='TF MOtif Discovery from Importance SCOres',
           long_description="""Algorithm for discovering consolidated patterns from base-pair-level importance scores""",
           url='https://github.com/kundajelab/tfmodisco',
-          version='0.5.0.0',
+          version='0.5.0.1',
           packages=find_packages(),
           package_data={
                 '': ['cluster/phenograph/louvain/*convert*', 'cluster/phenograph/louvain/*community*', 'cluster/phenograph/louvain/*hierarchy*']