phykit 2.1.32__tar.gz → 2.1.34__tar.gz

{phykit-2.1.32 → phykit-2.1.34}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phykit
-Version: 2.1.32
+Version: 2.1.34
 Home-page: https://github.com/jlsteenwyk/phykit
 Author: Jacob L. Steenwyk
 Author-email: jlsteenwyk@gmail.com

{phykit-2.1.32 → phykit-2.1.34}/phykit/services/tree/discordance_asymmetry.py

@@ -32,7 +32,7 @@ class DiscordanceAsymmetry(Tree):
         species_tree = self.read_tree_file()
         gene_trees = self._parse_gene_trees(self.gene_trees_path)
 
-        topology_counts = self._count_topologies(species_tree, gene_trees)
+        topology_counts, shared_taxa = self._count_topologies(species_tree, gene_trees)
 
         # Test each branch and collect results
         branch_results = []
@@ -82,7 +82,8 @@ class DiscordanceAsymmetry(Tree):
             self._output_text(branch_results, summary)
 
         if self.plot_output:
-            self._plot(species_tree, branch_results, self.plot_output)
+            self._plot(species_tree, branch_results, self.plot_output,
+                       shared_taxa=shared_taxa)
 
     # ------------------------------------------------------------------
     # Output methods
@@ -162,7 +163,8 @@ class DiscordanceAsymmetry(Tree):
         )
         print_json(result)
 
-    def _plot(self, species_tree, branch_results, output_path) -> None:
+    def _plot(self, species_tree, branch_results, output_path,
+              shared_taxa=None) -> None:
         """Phylogram colored by asymmetry ratio at each branch."""
         import matplotlib
         matplotlib.use("Agg")
@@ -178,7 +180,10 @@ class DiscordanceAsymmetry(Tree):
 
         parent_map = self._build_parent_map(species_tree)
         tips = list(species_tree.get_terminals())
-        all_taxa_fs = frozenset(t.name for t in tips)
+        # Use shared taxa (intersection with gene trees) for split label
+        # matching so labels are consistent with _count_topologies.
+        all_taxa_fs = (shared_taxa if shared_taxa is not None
+                       else frozenset(t.name for t in tips))
 
         # Compute node positions
         node_x = {}
@@ -212,7 +217,7 @@ class DiscordanceAsymmetry(Tree):
         for clade in species_tree.find_clades(order="preorder"):
             if clade.is_terminal():
                 continue
-            node_tips = frozenset(t.name for t in clade.get_terminals())
+            node_tips = frozenset(t.name for t in clade.get_terminals()) & all_taxa_fs
             split_label = (
                 sorted(node_tips)
                 if len(node_tips) <= len(all_taxa_fs) - len(node_tips)
@@ -361,6 +366,51 @@ class DiscordanceAsymmetry(Tree):
                 parent_map[id(child)] = clade
         return parent_map
 
+    @staticmethod
+    def _collect_taxa(trees) -> set:
+        """Collect the union of all tip names across a list of trees.
+
+        Uses a direct stack walk over .clades to avoid the overhead of
+        Bio.Phylo's find_clades / match_attrs dispatch layer.
+        """
+        taxa = set()
+        for tree in trees:
+            stack = [tree.root]
+            while stack:
+                node = stack.pop()
+                if not node.clades:
+                    taxa.add(node.name)
+                else:
+                    stack.extend(node.clades)
+        return taxa
+
+    def _extract_splits(self, tree, all_taxa_fs) -> set:
+        """Extract canonical bipartitions from a single tree.
+
+        Performs a single postorder traversal, building tip sets
+        bottom-up via .clades, and canonicalises each non-trivial split.
+        """
+        splits = set()
+        tip_sets = {}
+        stack = [(tree.root, False)]
+        while stack:
+            node, children_done = stack[-1]
+            if not node.clades:
+                stack.pop()
+                name = node.name
+                tip_sets[id(node)] = frozenset((name,)) if name in all_taxa_fs else frozenset()
+            elif children_done:
+                stack.pop()
+                merged = frozenset().union(*(tip_sets[id(c)] for c in node.clades))
+                tip_sets[id(node)] = merged
+                if len(merged) > 1 and merged != all_taxa_fs:
+                    splits.add(self._canonical_split(merged, all_taxa_fs))
+            else:
+                stack[-1] = (node, True)
+                for child in reversed(node.clades):
+                    stack.append((child, False))
+        return splits
+
     def _get_four_groups(self, tree, node, parent_map, all_taxa_fs):
         """Identify the four subtree groups around an internal branch.
 
@@ -376,26 +426,54 @@ class DiscordanceAsymmetry(Tree):
         if node.is_terminal() or len(node.clades) < 2:
             return None
 
-        C1 = frozenset(t.name for t in node.clades[0].get_terminals())
-        C2 = frozenset(t.name for t in node.clades[1].get_terminals())
+        C1 = frozenset(t.name for t in node.clades[0].get_terminals()) & all_taxa_fs
+        C2 = frozenset(t.name for t in node.clades[1].get_terminals()) & all_taxa_fs
         # If node has >2 children (polytomy), merge extras into C2
         for extra_child in node.clades[2:]:
-            C2 = C2 | frozenset(t.name for t in extra_child.get_terminals())
+            C2 = C2 | (frozenset(t.name for t in extra_child.get_terminals()) & all_taxa_fs)
 
         parent = parent_map.get(id(node))
         if parent is None:
             # node is root — no branch above it
             return None
 
-        # Get siblings of node under parent
+        # Get siblings of node under parent; pick the first sibling
+        # whose shared taxa are non-empty (avoids skipping valid branches
+        # when the first sibling's taxa are all absent from gene trees).
         siblings = [c for c in parent.clades if id(c) != id(node)]
         if not siblings:
             return None
 
-        S = frozenset(t.name for t in siblings[0].get_terminals())
+        chosen_sib = None
+        S = frozenset()
+        for sib in siblings:
+            candidate = frozenset(t.name for t in sib.get_terminals()) & all_taxa_fs
+            if candidate:
+                S = candidate
+                chosen_sib = sib
+                break
+
+        if not S:
+            return None
+
         # D = everything else (other siblings + above parent)
         D = all_taxa_fs - C1 - C2 - S
 
+        # For the root branch (D empty), the sibling encompasses the
+        # entire other side of the tree.  Decompose the sibling into its
+        # children to obtain proper 4-subtree NNI groups.
+        if not D:
+            if chosen_sib.is_terminal() or len(chosen_sib.clades) < 2:
+                return None
+            S = frozenset(t.name for t in chosen_sib.clades[0].get_terminals()) & all_taxa_fs
+            D = frozenset(t.name for t in chosen_sib.clades[1].get_terminals()) & all_taxa_fs
+            for extra in chosen_sib.clades[2:]:
+                D = D | (frozenset(t.name for t in extra.get_terminals()) & all_taxa_fs)
+
+        # Skip if any group is empty (branch is degenerate after taxon filtering)
+        if not C1 or not C2 or not S:
+            return None
+
         return C1, C2, S, D
 
     def _count_topologies(self, species_tree, gene_trees) -> Dict:
@@ -409,28 +487,17 @@ class DiscordanceAsymmetry(Tree):
           n_alt1: int
           n_alt2: int
         """
-        all_taxa = sorted(
-            set(t.name for t in species_tree.get_terminals())
-            & set().union(*(
-                set(t.name for t in gt.get_terminals()) for gt in gene_trees
-            ))
-        )
-        all_taxa_fs = frozenset(all_taxa)
+        species_taxa = set(t.name for t in species_tree.get_terminals())
+        gene_taxa = self._collect_taxa(gene_trees)
+        all_taxa_fs = frozenset(sorted(species_taxa & gene_taxa))
         parent_map = self._build_parent_map(species_tree)
 
         # Extract bipartitions from all gene trees (topology only, no lengths).
-        # Restrict bipartitions to shared taxa without mutating gene tree objects.
+        # Builds tip sets bottom-up in a single postorder pass per gene tree,
+        # avoiding repeated get_terminals() calls (O(n) vs O(n²) per tree).
         gene_tree_splits = []
         for gt in gene_trees:
-            splits = set()
-            for clade in gt.get_nonterminals():
-                tips = frozenset(
-                    t.name for t in clade.get_terminals()
-                    if t.name in all_taxa_fs
-                )
-                if len(tips) <= 1 or tips == all_taxa_fs:
-                    continue
-                splits.add(self._canonical_split(tips, all_taxa_fs))
+            splits = self._extract_splits(gt, all_taxa_fs)
             gene_tree_splits.append(splits)
 
         result = {}
@@ -452,7 +519,7 @@ class DiscordanceAsymmetry(Tree):
             n_alt1 = sum(1 for splits in gene_tree_splits if nni_alt1_bp in splits)
             n_alt2 = sum(1 for splits in gene_tree_splits if nni_alt2_bp in splits)
 
-            node_tips = frozenset(t.name for t in clade.get_terminals())
+            node_tips = frozenset(t.name for t in clade.get_terminals()) & all_taxa_fs
             split_label = (
                 sorted(node_tips)
                 if len(node_tips) <= len(all_taxa_fs) - len(node_tips)
@@ -465,7 +532,7 @@ class DiscordanceAsymmetry(Tree):
                 n_alt1=n_alt1,
                 n_alt2=n_alt2,
             )
-        return result
+        return result, all_taxa_fs
 
     # ------------------------------------------------------------------
     # Statistical testing

phykit-2.1.34/phykit/version.py

@@ -0,0 +1 @@
+__version__ = "2.1.34"

{phykit-2.1.32 → phykit-2.1.34}/phykit.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phykit
-Version: 2.1.32
+Version: 2.1.34
 Home-page: https://github.com/jlsteenwyk/phykit
 Author: Jacob L. Steenwyk
 Author-email: jlsteenwyk@gmail.com

phykit-2.1.32/phykit/version.py

@@ -1 +0,0 @@
-__version__ = "2.1.32"