Merge pull request #70 from kundajelab/fixedgecasesigndisagreement

Retrying what was supposed to be PR 67

.travis.yml

@@ -5,6 +5,8 @@ matrix:
     include:
         - python: 3.6
           env: TF_VERSION=1.14.0
+        - python: 3.8.3
+          env: TF_VERSION=2.3.0
 
 notifications:
   email: true

modisco.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: modisco
-Version: 0.5.8.0
+Version: 0.5.8.1
 Summary: TF MOtif Discovery from Importance SCOres
 Home-page: https://github.com/kundajelab/tfmodisco
 License: UNKNOWN

modisco/affinitymat/core.py

@@ -478,6 +478,7 @@ def __init__(self, n_cores, cross_metric_single_region, verbose=True):
         self.cross_metric_single_region = cross_metric_single_region
         self.verbose = verbose
 
+    #min_overlap is w.r.t. the length of 'filters'
     def __call__(self, filters, things_to_scan, min_overlap,
                        neighbors_of_things_to_scan=None,
                        return_sparse=False):

modisco/aggregator.py

@@ -276,7 +276,7 @@ def __call__(self, patterns):
                 large_patterns, new_assignments =\
                     self.seqlet_assigner(patterns=large_patterns,
                                          seqlets_to_assign=seqlets_to_assign,
-                                         merge_into_existing_patterns=True)
+                                         merge_into_existing_patterns=False)
             large_patterns = self.postprocessor(large_patterns)
             return large_patterns
         else:
@@ -399,13 +399,16 @@ def __call__(self, patterns, seqlets_to_assign,
 
         if (merge_into_existing_patterns):
             new_patterns = patterns
-            for pattern,x in zip(patterns, seqlet_and_alnmnt_grps):
-                pattern.merge_seqlets_and_alnmts(
-                    seqlets_and_alnmts=x,
-                    aligner=self.pattern_aligner) 
         else:
-            new_patterns = [core.AggregatedSeqlet(seqlets_and_alnmts_arr=x)
-                for x in seqlet_and_alnmnt_grps if len(x) > 0]
+            #Make a copy of each one
+            new_patterns = [
+                core.AggregatedSeqlet(pattern._seqlets_and_alnmts.copy())
+                for pattern in patterns]
+
+        for pattern,x in zip(new_patterns, seqlet_and_alnmnt_grps):
+            pattern.merge_seqlets_and_alnmts(
+                seqlets_and_alnmts=x,
+                aligner=self.pattern_aligner) 
 
         return new_patterns, seqlet_assignments
 

modisco/backend/tensorflow_backend.py

@@ -1,5 +1,9 @@
 from __future__ import division, print_function
-import tensorflow as tf
+try:
+    import tensorflow.compat.v1 as tf
+    tf.disable_v2_behavior()
+except:
+    import tensorflow as tf
 import numpy as np
 import sys
 

modisco/tfmodisco_workflow/seqlets_to_patterns.py

@@ -507,11 +507,14 @@ class SeqletsToPatternsResults(object):
 
     def __init__(self,
                  each_round_initcluster_motifs, 
-                 patterns, cluster_results,
+                 patterns, 
+                 patterns_withoutreassignment,
+                 cluster_results,
                  total_time_taken, success=True, **kwargs):
         self.each_round_initcluster_motifs = each_round_initcluster_motifs
         self.success = success
         self.patterns = patterns
+        self.patterns_withoutreassignment = patterns_withoutreassignment
         self.cluster_results = cluster_results
         self.total_time_taken = total_time_taken
         self.__dict__.update(**kwargs)
@@ -553,11 +556,18 @@ def from_hdf5(cls, grp, track_set):
                         track_set=track_set)
             patterns = util.load_patterns(grp=grp["patterns"],
                                           track_set=track_set) 
+            if "patterns_withoutreassignment" in grp:
+                patterns_withoutreassignment = util.load_patterns(
+                    grp=grp["patterns_withoutreassignment"],
+                    track_set=track_set) 
+            else: #backwards compatibility
+                patterns_withoutreassignment = []
             cluster_results = None
             total_time_taken = None
             return cls(
                 each_round_initcluster_motifs=each_round_initcluster_motifs,
                 patterns=patterns,
+                patterns_withoutreassignment=patterns_withoutreassignment,
                 cluster_results=cluster_results,
                 total_time_taken=total_time_taken)
         else:
@@ -572,6 +582,9 @@ def save_hdf5(self, grp):
                     grp=grp.create_group("each_round_initcluster_motifs"))
             util.save_patterns(self.patterns,
                                grp.create_group("patterns"))
+            util.save_patterns(
+                self.patterns_withoutreassignment,
+                grp.create_group("patterns_withoutreassignment"))
             self.cluster_results.save_hdf5(grp.create_group("cluster_results"))   
             grp.attrs['total_time_taken'] = self.total_time_taken
 
@@ -656,7 +669,6 @@ def get_cluster_to_aggregate_motif(self, seqlets, cluster_indices,
                                   " with "+str(motif.num_seqlets)
                                   +" seqlets due to sign disagreement")
                         cluster_to_eliminated_motif[i] = motif
-                cluster_to_motif[i] = motif
         return cluster_to_motif, cluster_to_eliminated_motif
 
 
@@ -892,6 +904,8 @@ def __call__(self, seqlets):
             sys.stdout.flush()
         reassigned_patterns = self.seqlet_reassigner(merged_patterns)
         final_patterns = self.final_postprocessor(reassigned_patterns)
+        final_patterns_withoutreassignment =\
+            self.final_postprocessor(merged_patterns)
         if (self.verbose):
             print("Got "+str(len(final_patterns))
                   +" patterns after reassignment")
@@ -908,6 +922,7 @@ def __call__(self, seqlets):
         results = SeqletsToPatternsResults(
             each_round_initcluster_motifs=each_round_initcluster_motifs,             
             patterns=final_patterns,
+            patterns_withoutreassignment=final_patterns_withoutreassignment,
             seqlets=filtered_seqlets, #last stage of filtered seqlets
             #affmat=filtered_affmat,
             cluster_results=cluster_results, 

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='0.5.8.0',
+          version='0.5.8.1',
           packages=find_packages(),
           package_data={
                 '': ['cluster/phenograph/louvain/*convert*', 'cluster/phenograph/louvain/*community*', 'cluster/phenograph/louvain/*hierarchy*']