updated sorting in recursive_alignement clusters

baliga-lab · Jul 1, 2024 · d266539 · d266539
1 parent 065b500
commit d266539
Show file tree

Hide file tree

Showing 3 changed files with 9 additions and 7 deletions.
diff --git a/miner/miner.py b/miner/miner.py
@@ -820,7 +820,7 @@ def remix(df,frequencyClusters):
         minGenes = max(4,cut*len(cluster))
         keepers = list(sliceDf.columns[np.where(sumSlice>=minGenes)[0]])
         keepers = list(set(keepers)|set(cluster))
-        finalClusters.append(keepers)
+        finalClusters.append(sorted(keepers))
         finalClusters.sort(key = lambda s: -len(s))
     return finalClusters
 
@@ -951,7 +951,7 @@ def combineClusters(axes,clusters,threshold=0.925):
     revisedClusters = {}
     combinedKeys = decomposeDictionaryToLists(combineAxes)
     for keyList in combinedKeys:
-        genes = list(set(np.hstack([clusters[i] for i in keyList])))
+        genes = sorted(set(np.hstack([clusters[i] for i in keyList])))
         revisedClusters[len(revisedClusters)] = genes
 
     return revisedClusters
@@ -966,8 +966,8 @@ def reconstruction(decomposedList,expressionData, random_state, threshold=0.925)
 
     clusters = {i:decomposedList[i] for i in range(len(decomposedList))}
     axes = get_axes(clusters, expressionData, random_state)
-    recombine = combineClusters(axes,clusters,threshold)
-    return recombine
+    return combineClusters(axes,clusters,threshold)
+
 
 def recursive_alignment(geneset,expressionData,minNumberGenes=6,
                         pct_threshold=80, random_state=12):
@@ -976,7 +976,7 @@ def recursive_alignment(geneset,expressionData,minNumberGenes=6,
         return []
 
     reconstructed = reconstruction(recDecomp,expressionData, random_state)
-    reconstructedList = [reconstructed[i] for i in list(reconstructed.keys()) if len(reconstructed[i])>minNumberGenes]
+    reconstructedList = [reconstructed[i] for i in list(reconstructed.keys()) if len(reconstructed[i]) > minNumberGenes]
     reconstructedList.sort(key = lambda s: -len(s))
     return reconstructedList
 

diff --git a/test/mechinf_test.py b/test/mechinf_test.py
@@ -84,7 +84,9 @@ def test_recursive_alignment():
             ref_recalign.append(line.strip().split(" "))
 
     rec_align = miner.recursive_alignment(cluster, exp, 6, 80)
-    assert(ref_recalign == rec_align)
+    #with open("ref_recalign.txt", "w") as outfile:
+    #    outfile.write("%s" % ' '.join(sorted(rec_align[0])))
+    assert(ref_recalign[0] == rec_align[0])
 
 
 def test_revise_initial_clusters():

diff --git a/testdata/ref_recalign.txt b/testdata/ref_recalign.txt
@@ -1 +1 @@
-ENSG00000006756 ENSG00000040487 ENSG00000101605 ENSG00000175556 ENSG00000179820 ENSG00000198799 ENSG00000166484 ENSG00000019995 ENSG00000092148
+ENSG00000006756 ENSG00000019995 ENSG00000040487 ENSG00000092148 ENSG00000101605 ENSG00000166484 ENSG00000175556 ENSG00000179820 ENSG00000198799