phykit 2.1.84__tar.gz → 2.1.85__tar.gz

{phykit-2.1.84 → phykit-2.1.85}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: phykit
-Version: 2.1.84
+Version: 2.1.85
 Home-page: https://github.com/jlsteenwyk/phykit
 Author: Jacob L. Steenwyk
 Author-email: jlsteenwyk@gmail.com

{phykit-2.1.84 → phykit-2.1.85}/phykit/cli_registry.py

@@ -215,6 +215,9 @@ ALIAS_TO_HANDLER: Dict[str, str] = {
     "tree_space": "tree_space",
     "tspace": "tree_space",
     "tree_landscape": "tree_space",
+    "transfer_annotations": "transfer_annotations",
+    "transfer_annot": "transfer_annotations",
+    "annotate_tree": "transfer_annotations",
     "tgroups": "taxon_groups",
     "shared_taxa": "taxon_groups",
     "occupancy_filter": "occupancy_filter",

{phykit-2.1.84 → phykit-2.1.85}/phykit/phykit.py

@@ -308,6 +308,8 @@ class Phykit:
                     - prune taxa from a phylogeny
                 subtree_prune_regraft (alias: spr)
                     - generate all SPR rearrangements for a specified subtree
+                transfer_annotations (alias: transfer_annot; annotate_tree)
+                    - transfer node annotations between trees (e.g., wASTRAL to RAxML/IQ-TREE)
                 relative_rate_test (alias: rrt; tajima_rrt)
                     - Tajima's relative rate test for equal evolutionary
                       rates between two ingroup lineages
@@ -6821,6 +6823,59 @@ class Phykit:
         _add_json_argument(parser)
         _run_service(parser, argv, Spr)
 
+    @staticmethod
+    def transfer_annotations(argv):
+        parser = _new_parser(
+            description=textwrap.dedent(
+                f"""\
+                {help_header}
+
+                Transfer internal node annotations from one tree onto
+                another. Matches nodes by bipartition (descendant taxa
+                set) and copies the annotation labels.
+
+                Typical use case: transfer wASTRAL support annotations
+                (q1/q2/q3, pp1, f1, etc.) from an annotated ASTRAL
+                tree onto a branch-length-optimized topology from
+                RAxML-NG, IQ-TREE, or any other tool. The output tree
+                has the target's branch lengths with the source's
+                annotations.
+
+                Aliases:
+                  transfer_annotations, transfer_annot, annotate_tree
+                Command line interfaces:
+                  pk_transfer_annotations, pk_transfer_annot, pk_annotate_tree
+
+                Usage:
+                phykit transfer_annotations --source <annotated_tree>
+                  --target <branch_length_tree> [-o/--output <file>]
+                  [--json]
+
+                Options
+                =====================================================
+                --source                    annotated tree file (e.g.,
+                                            wASTRAL output with
+                                            --support 3)
+
+                --target                    target tree file with
+                                            branch lengths to keep
+                                            (e.g., RAxML-NG or
+                                            IQ-TREE output)
+
+                -o/--output                 output file for the
+                                            annotated tree (default:
+                                            target file + ".annotated")
+
+                --json                      output results as JSON
+                """
+            ),
+        )
+        parser.add_argument("--source", type=str, required=True, help=SUPPRESS, metavar="")
+        parser.add_argument("--target", type=str, required=True, help=SUPPRESS, metavar="")
+        parser.add_argument("-o", "--output", type=str, default=None, help=SUPPRESS, metavar="")
+        _add_json_argument(parser)
+        _run_service(parser, argv, TransferAnnotations)
+
     @staticmethod
     def relative_rate_test(argv):
         parser = _new_parser(
@@ -9190,6 +9245,10 @@ def subtree_prune_regraft(argv=None):
     Phykit.subtree_prune_regraft(sys.argv[1:])
 
 
+def transfer_annotations(argv=None):
+    Phykit.transfer_annotations(sys.argv[1:])
+
+
 def relative_rate_test(argv=None):
     Phykit.relative_rate_test(sys.argv[1:])
 

{phykit-2.1.84 → phykit-2.1.85}/phykit/service_factories.py

@@ -108,6 +108,7 @@ RobinsonFouldsDistance = _LazyServiceFactory("phykit.services.tree.rf_distance",
 RootTree = _LazyServiceFactory("phykit.services.tree.root_tree", "RootTree")
 Saturation = _LazyServiceFactory("phykit.services.tree.saturation", "Saturation")
 Spr = _LazyServiceFactory("phykit.services.tree.spr", "Spr")
+TransferAnnotations = _LazyServiceFactory("phykit.services.tree.transfer_annotations", "TransferAnnotations")
 SpuriousSequence = _LazyServiceFactory("phykit.services.tree.spurious_sequence", "SpuriousSequence")
 TerminalBranchStats = _LazyServiceFactory("phykit.services.tree.terminal_branch_stats", "TerminalBranchStats")
 TipLabels = _LazyServiceFactory("phykit.services.tree.tip_labels", "TipLabels")

{phykit-2.1.84 → phykit-2.1.85}/phykit/services/tree/__init__.py

@@ -48,6 +48,7 @@ _EXPORTS = {
     "TipToTipDistance": "tip_to_tip_distance",
     "TipToTipNodeDistance": "tip_to_tip_node_distance",
     "TotalTreeLength": "total_tree_length",
+    "TransferAnnotations": "transfer_annotations",
     "Treeness": "treeness",
     "ThresholdModel": "threshold_model",
     "TreenessOverRCV": "treeness_over_rcv",

phykit-2.1.85/phykit/services/tree/transfer_annotations.py

@@ -0,0 +1,166 @@
+"""
+Transfer annotations between phylogenies.
+
+Copies internal node annotations (e.g., wASTRAL q1/q2/q3, pp1, f1)
+from an annotated source tree onto a target tree with optimized branch
+lengths. Nodes are matched by their bipartition (set of descendant taxa).
+
+Typical use case: transfer wASTRAL support annotations onto a
+branch-length-optimized topology from RAxML-NG, IQ-TREE, or any
+other tool for use with quartet_pie or other visualization commands.
+"""
+from io import StringIO
+from typing import Dict, List, Set, Tuple
+
+from Bio import Phylo
+
+from .base import Tree
+from ...helpers.json_output import print_json
+from ...errors import PhykitUserError
+
+
+class TransferAnnotations(Tree):
+    def __init__(self, args) -> None:
+        parsed = self.process_args(args)
+        super().__init__(tree_file_path=parsed["target_path"])
+        self.source_path = parsed["source_path"]
+        self.output_path = parsed["output_path"]
+        self.json_output = parsed["json_output"]
+
+    def process_args(self, args) -> Dict:
+        return dict(
+            target_path=args.target,
+            source_path=args.source,
+            output_path=getattr(args, "output", None),
+            json_output=getattr(args, "json", False),
+        )
+
+    def run(self) -> None:
+        # Read both trees
+        target = self.read_tree_file()
+        try:
+            source = Phylo.read(self.source_path, "newick")
+        except FileNotFoundError:
+            raise PhykitUserError(
+                [
+                    f"{self.source_path} corresponds to no such file or directory.",
+                    "Please check filename and pathing",
+                ],
+                code=2,
+            )
+
+        # Validate taxa match
+        target_tips = {t.name for t in target.get_terminals()}
+        source_tips = {t.name for t in source.get_terminals()}
+
+        if target_tips != source_tips:
+            only_target = target_tips - source_tips
+            only_source = source_tips - target_tips
+            msgs = ["Taxa in source and target trees do not match."]
+            if only_target:
+                msgs.append(
+                    f"Only in target: {', '.join(sorted(only_target)[:5])}"
+                    + (f" ... ({len(only_target)} total)" if len(only_target) > 5 else "")
+                )
+            if only_source:
+                msgs.append(
+                    f"Only in source: {', '.join(sorted(only_source)[:5])}"
+                    + (f" ... ({len(only_source)} total)" if len(only_source) > 5 else "")
+                )
+            raise PhykitUserError(msgs, code=2)
+
+        # Build bipartition -> annotation map from source tree
+        all_taxa = frozenset(source_tips)
+        source_annotations = self._extract_annotations(source, all_taxa)
+
+        # Match and transfer annotations to target tree
+        n_transferred, n_unmatched = self._transfer(
+            target, source_annotations, all_taxa
+        )
+
+        # Output
+        output_path = self.output_path or f"{self.tree_file_path}.annotated"
+        self._write_annotated_tree(target, output_path)
+
+        if self.json_output:
+            self._print_json(
+                n_transferred, n_unmatched, output_path,
+                len(source_annotations), target_tips,
+            )
+        else:
+            try:
+                print(f"Annotations transferred: {n_transferred}")
+                print(f"Unmatched nodes: {n_unmatched}")
+                print(f"Output: {output_path}")
+            except BrokenPipeError:
+                return
+
+    @staticmethod
+    def _get_bipartition(clade, all_taxa: frozenset) -> frozenset:
+        """Get canonical bipartition for a clade (smaller side)."""
+        tips = frozenset(t.name for t in clade.get_terminals())
+        complement = all_taxa - tips
+        # Use the smaller side as the canonical key
+        return tips if len(tips) <= len(complement) else complement
+
+    def _extract_annotations(
+        self, tree, all_taxa: frozenset
+    ) -> Dict[frozenset, str]:
+        """Extract bipartition -> annotation mapping from source tree."""
+        annotations = {}
+        for clade in tree.find_clades(order="preorder"):
+            if clade.is_terminal() or clade == tree.root:
+                continue
+            annotation = clade.comment or clade.name or ""
+            if not annotation:
+                continue
+            bp = self._get_bipartition(clade, all_taxa)
+            if bp:
+                annotations[bp] = annotation
+        return annotations
+
+    def _transfer(
+        self, target, source_annotations: Dict[frozenset, str],
+        all_taxa: frozenset,
+    ) -> Tuple[int, int]:
+        """Transfer annotations from source to target by bipartition matching."""
+        transferred = 0
+        unmatched = 0
+
+        for clade in target.find_clades(order="preorder"):
+            if clade.is_terminal() or clade == target.root:
+                continue
+            bp = self._get_bipartition(clade, all_taxa)
+            if bp in source_annotations:
+                clade.comment = source_annotations[bp]
+                transferred += 1
+            else:
+                unmatched += 1
+
+        return transferred, unmatched
+
+    @staticmethod
+    def _write_annotated_tree(tree, output_path: str) -> None:
+        """Write tree preserving comment annotations in brackets."""
+        Phylo.write(tree, output_path, "newick")
+        # BioPython writes comments as [&comment] but wASTRAL uses
+        # [comment]. Fix the format.
+        with open(output_path) as f:
+            content = f.read()
+        # Remove the & prefix that BioPython adds
+        content = content.replace("[&", "[")
+        with open(output_path, "w") as f:
+            f.write(content)
+
+    def _print_json(
+        self, n_transferred, n_unmatched, output_path,
+        n_source_annotations, target_tips,
+    ):
+        payload = {
+            "annotations_transferred": n_transferred,
+            "unmatched_nodes": n_unmatched,
+            "source_annotations_available": n_source_annotations,
+            "n_taxa": len(target_tips),
+            "output": output_path,
+        }
+        print_json(payload, sort_keys=False)

phykit-2.1.85/phykit/version.py

@@ -0,0 +1 @@
+__version__ = "2.1.85"

{phykit-2.1.84 → phykit-2.1.85}/phykit.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: phykit
-Version: 2.1.84
+Version: 2.1.85
 Home-page: https://github.com/jlsteenwyk/phykit
 Author: Jacob L. Steenwyk
 Author-email: jlsteenwyk@gmail.com

{phykit-2.1.84 → phykit-2.1.85}/phykit.egg-info/SOURCES.txt

@@ -136,6 +136,7 @@ phykit/services/tree/tip_to_tip_node_distance.py
 phykit/services/tree/total_tree_length.py
 phykit/services/tree/trait_correlation.py
 phykit/services/tree/trait_rate_map.py
+phykit/services/tree/transfer_annotations.py
 phykit/services/tree/tree_space.py
 phykit/services/tree/treeness.py
 phykit/services/tree/treeness_over_rcv.py

{phykit-2.1.84 → phykit-2.1.85}/phykit.egg-info/entry_points.txt

@@ -9,6 +9,7 @@ pk_aln_recoding = phykit.phykit:alignment_recoding
 pk_aln_subsample = phykit.phykit:alignment_subsample
 pk_alng = phykit.phykit:alignment_length_no_gaps
 pk_anc_recon = phykit.phykit:ancestral_state_reconstruction
+pk_annotate_tree = phykit.phykit:transfer_annotations
 pk_aot = phykit.phykit:alignment_outlier_taxa
 pk_asr = phykit.phykit:ancestral_state_reconstruction
 pk_blm = phykit.phykit:branch_length_multiplier
@@ -207,6 +208,8 @@ pk_trait_corr = phykit.phykit:trait_correlation
 pk_trait_correlation = phykit.phykit:trait_correlation
 pk_trait_rate_map = phykit.phykit:trait_rate_map
 pk_traitgram = phykit.phykit:phenogram
+pk_transfer_annot = phykit.phykit:transfer_annotations
+pk_transfer_annotations = phykit.phykit:transfer_annotations
 pk_tree_labels = phykit.phykit:tip_labels
 pk_tree_landscape = phykit.phykit:tree_space
 pk_tree_len = phykit.phykit:total_tree_length

phykit-2.1.84/phykit/version.py

@@ -1 +0,0 @@
-__version__ = "2.1.84"