should be cartesian product instead of combination (#312)

* should be cartesian product

* Update _tools.py

* Update _plotting.py

* Update _plotting.py

* Update 4_dandelion_visualization-10x_data.ipynb

dandelion/plotting/_plotting.py

@@ -6,7 +6,7 @@
 import seaborn as sns
 
 from anndata import AnnData
-from itertools import combinations, cycle
+from itertools import product, cycle
 from matplotlib.axes import Axes
 from matplotlib.figure import Figure
 import networkx as nx
@@ -929,23 +929,24 @@ def clone_overlap(
                 edges[x] = [
                     y + ({str(clone_): x},)
                     for y in list(
-                        combinations(
+                        product(
                             [
                                 i
                                 for i in overlap.loc[x][
                                     overlap.loc[x] > 0
                                 ].index
                             ],
-                            2,
+                            repeat=2,
                         )
                     )
                 ]
     else:
         tmp_overlap = overlap.astype(bool).sum(axis=1)
         combis = {
             x: list(
-                combinations(
-                    [i for i in overlap.loc[x][overlap.loc[x] > 0].index], 2
+                product(
+                    [i for i in overlap.loc[x][overlap.loc[x] > 0].index],
+                    repeat=2,
                 )
             )
             for x in tmp_overlap.index
@@ -972,18 +973,20 @@ def clone_overlap(
                     + (
                         {
                             str(clone_): x,
-                            "weight": tmp_edge_weight_dict[y],
+                            "weight": tmp_edge_weight_dict[y]
+                            if not isinstance(tmp_edge_weight_dict[y], list)
+                            else 0,
                         },
                     )
                     for y in list(
-                        combinations(
+                        product(
                             [
                                 i
                                 for i in overlap.loc[x][
                                     overlap.loc[x] > 0
                                 ].index
                             ],
-                            2,
+                            repeat=2,
                         )
                     )
                 ]
@@ -1064,6 +1067,8 @@ def clone_overlap(
         if return_heatmap_data:
             return hm
     else:
+        # remove self loops
+        G.remove_edges_from(nx.selfloop_edges(G))
         ax = nxv.circos(
             G,
             group_by=colorby,

dandelion/tools/_tools.py

@@ -13,7 +13,7 @@
 from changeo.Gene import getGene
 from collections import defaultdict, Counter
 from distance import hamming
-from itertools import combinations
+from itertools import product
 from scanpy import logging as logg
 from scipy.sparse import csr_matrix
 from scipy.spatial.distance import pdist, squareform
@@ -1675,19 +1675,19 @@ def clustering(distance_dict, threshold, sequences_dict):
         (i1, i2): distance_dict[(i1, i2)] <= threshold
         if (i1, i2) in distance_dict
         else False
-        for i1, i2 in combinations(i_unique, 2)
+        for i1, i2 in product(i_unique, repeat=2)
     }
     i_pair_d.update(
         {
             (i2, i1): distance_dict[(i2, i1)] <= threshold
             if (i2, i1) in distance_dict
             else False
-            for i1, i2 in combinations(i_unique, 2)
+            for i1, i2 in product(i_unique, repeat=2)
         }
     )
     # so which indices should not be part of a clone?
     canbetogether = defaultdict(list)
-    for ii1, ii2 in combinations(i_unique, 2):
+    for ii1, ii2 in product(i_unique, repeat=2):
         if i_pair_d[(ii1, ii2)] or i_pair_d[(ii2, ii1)]:
             if (ii1, ii2) in distance_dict:
                 canbetogether[ii1].append((ii1, ii2))