phykit 2.1.79__tar.gz → 2.1.81__tar.gz

{phykit-2.1.79 → phykit-2.1.81}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: phykit
-Version: 2.1.79
+Version: 2.1.81
 Home-page: https://github.com/jlsteenwyk/phykit
 Author: Jacob L. Steenwyk
 Author-email: jlsteenwyk@gmail.com

{phykit-2.1.79 → phykit-2.1.81}/phykit/cli_registry.py

@@ -55,6 +55,10 @@ ALIAS_TO_HANDLER: Dict[str, str] = {
     "sum_of_pairs_score": "sum_of_pairs_score",
     "sops": "sum_of_pairs_score",
     "sop": "sum_of_pairs_score",
+    "phylo_anova": "phylo_anova",
+    "panova": "phylo_anova",
+    "phylo_manova": "phylo_anova",
+    "pmanova": "phylo_anova",
     "phylo_gwas": "phylo_gwas",
     "pgwas": "phylo_gwas",
     "vs": "variable_sites",

{phykit-2.1.79 → phykit-2.1.81}/phykit/helpers/quartet_utils.py

@@ -114,6 +114,42 @@ def parse_astral_annotations(
     return result
 
 
+def parse_astral_branch_info(
+    tree,
+) -> Dict[int, Dict[str, float]]:
+    """Parse f1 (concordant count) and pp1 (LPP) from ASTRAL/wASTRAL labels.
+
+    Returns dict mapping clade id -> {"f1": ..., "pp1": ...}.
+    Keys are only present if the value was found in the annotation.
+    """
+    result = {}
+    for clade in tree.find_clades(order="preorder"):
+        if clade.is_terminal():
+            continue
+        label = clade.name or clade.comment or ""
+        info = _parse_branch_info_from_label(str(label))
+        if info:
+            result[id(clade)] = info
+    return result
+
+
+def _parse_branch_info_from_label(label: str) -> Dict[str, float]:
+    """Extract f1, pp1 (and optionally f2, f3, pp2, pp3) from an ASTRAL label."""
+    s = label.strip("'\"").strip("[]")
+    info = {}
+    for part in s.split(";"):
+        if "=" not in part:
+            continue
+        key, val = part.split("=", 1)
+        key = key.strip()
+        try:
+            if key in ("f1", "f2", "f3", "pp1", "pp2", "pp3"):
+                info[key] = float(val)
+        except ValueError:
+            continue
+    return info
+
+
 def _parse_qs_from_label(label: str) -> Optional[Tuple[float, float, float]]:
     """Extract q1, q2, q3 from an ASTRAL node label string."""
     s = label.strip("'\"").strip("[]")

{phykit-2.1.79 → phykit-2.1.81}/phykit/phykit.py

@@ -248,6 +248,8 @@ class Phykit:
                     - fit phylogenetic generalized least squares (PGLS) regression
                 phylogenetic_glm (alias: phylo_glm; pglm)
                     - fit phylogenetic GLM for binary (logistic) or count (Poisson) data
+                phylo_anova (alias: panova; phylo_manova; pmanova)
+                    - phylogenetic ANOVA / MANOVA using RRPP (Adams & Collyer 2018)
                 phylo_logistic (alias: phylo_logreg; plogreg)
                     - fit phylogenetic logistic regression (Ives & Garland 2010)
                 stochastic_character_map (alias: simmap; scm)
@@ -2004,6 +2006,87 @@ class Phykit:
         _add_json_argument(parser)
         _run_service(parser, argv, PhyloGwas)
 
+    @staticmethod
+    def phylo_anova(argv):
+        parser = _new_parser(
+            description=textwrap.dedent(
+                f"""\
+                {help_header}
+
+                Phylogenetic ANOVA / MANOVA using the Residual
+                Randomization Permutation Procedure (RRPP) of
+                Adams & Collyer (2018).
+
+                Tests whether a continuous trait (ANOVA) or multiple
+                traits (MANOVA) differ across discrete groups while
+                accounting for phylogenetic non-independence.
+
+                Auto-detects univariate vs multivariate based on the
+                number of response trait columns. Override with
+                --method anova or --method manova.
+
+                Aliases:
+                  phylo_anova, panova, phylo_manova, pmanova
+                Command line interfaces:
+                  pk_phylo_anova, pk_panova, pk_phylo_manova, pk_pmanova
+
+                Usage:
+                phykit phylo_anova -t <tree> --traits <traits_file>
+                  [--group-column <name>] [--method auto|anova|manova]
+                  [--permutations <int>] [--pairwise]
+                  [--plot-output <file>] [--plot-type boxplot|phylomorphospace]
+                  [--seed <int>] [--json]
+
+                Options
+                =====================================================
+                -t/--tree                   species tree file (required)
+
+                --traits                    TSV file with taxon, group
+                                            column, and one or more
+                                            response trait columns
+                                            (required)
+
+                --group-column              name of the categorical
+                                            grouping column (default:
+                                            first non-taxon column)
+
+                --method                    analysis method: auto, anova,
+                                            or manova (default: auto)
+
+                --permutations              number of RRPP permutations
+                                            (default: 1000)
+
+                --pairwise                  include post-hoc pairwise
+                                            group comparisons
+
+                --plot-output               output figure path
+                                            (.png, .pdf, .svg)
+
+                --plot-type                 boxplot or phylomorphospace
+                                            (default: auto — boxplot for
+                                            ANOVA, phylomorphospace for
+                                            MANOVA)
+
+                --seed                      random seed for reproducible
+                                            permutations
+
+                --json                      output results as JSON
+                """
+            ),
+        )
+        parser.add_argument("-t", "--tree", type=str, required=True, help=SUPPRESS, metavar="")
+        parser.add_argument("--traits", type=str, required=True, help=SUPPRESS, metavar="")
+        parser.add_argument("--group-column", type=str, default=None, help=SUPPRESS, metavar="")
+        parser.add_argument("--method", type=str, default="auto", choices=["auto", "anova", "manova"], help=SUPPRESS, metavar="")
+        parser.add_argument("--permutations", type=int, default=1000, help=SUPPRESS, metavar="")
+        parser.add_argument("--pairwise", action="store_true", help=SUPPRESS)
+        parser.add_argument("--plot-output", type=str, default=None, help=SUPPRESS, metavar="")
+        parser.add_argument("--plot-type", type=str, default="auto", choices=["auto", "boxplot", "phylomorphospace"], help=SUPPRESS, metavar="")
+        parser.add_argument("--seed", type=int, default=None, help=SUPPRESS, metavar="")
+        add_plot_arguments(parser)
+        _add_json_argument(parser)
+        _run_service(parser, argv, PhyloAnova)
+
     @staticmethod
     def alignment_subsample(argv):
         parser = _new_parser(
@@ -6076,6 +6159,11 @@ class Phykit:
                                             and branch colors (iTOL-
                                             inspired TSV format)
 
+                --branch-labels             show concordant gene count
+                                            above and LPP support below
+                                            each internal branch
+                                            (PhyTop-style)
+
                 --csv                       output per-branch concordance
                                             values as a CSV file
 
@@ -6102,6 +6190,9 @@ class Phykit:
         parser.add_argument(
             "--annotate", action="store_true", required=False, help=SUPPRESS
         )
+        parser.add_argument(
+            "--branch-labels", action="store_true", required=False, help=SUPPRESS
+        )
         parser.add_argument(
             "--csv", type=str, required=False, default=None,
             help=SUPPRESS, metavar=""
@@ -8677,6 +8768,10 @@ def phylo_gwas(argv=None):
     Phykit.phylo_gwas(sys.argv[1:])
 
 
+def phylo_anova(argv=None):
+    Phykit.phylo_anova(sys.argv[1:])
+
+
 def dfoil(argv=None):
     Phykit.dfoil(sys.argv[1:])
 

{phykit-2.1.79 → phykit-2.1.81}/phykit/service_factories.py

@@ -44,6 +44,7 @@ RelativeCompositionVariability = _LazyServiceFactory("phykit.services.alignment.
 RelativeCompositionVariabilityTaxon = _LazyServiceFactory("phykit.services.alignment.rcvt", "RelativeCompositionVariabilityTaxon")
 RenameFastaEntries = _LazyServiceFactory("phykit.services.alignment.rename_fasta_entries", "RenameFastaEntries")
 SumOfPairsScore = _LazyServiceFactory("phykit.services.alignment.sum_of_pairs_score", "SumOfPairsScore")
+PhyloAnova = _LazyServiceFactory("phykit.services.tree.phylo_anova", "PhyloAnova")
 PhyloGwas = _LazyServiceFactory("phykit.services.alignment.phylo_gwas", "PhyloGwas")
 TaxonGroups = _LazyServiceFactory("phykit.services.alignment.taxon_groups", "TaxonGroups")
 VariableSites = _LazyServiceFactory("phykit.services.alignment.variable_sites", "VariableSites")

{phykit-2.1.79 → phykit-2.1.81}/phykit/services/tree/__init__.py

@@ -52,6 +52,7 @@ _EXPORTS = {
     "TreenessOverRCV": "treeness_over_rcv",
     "ConcordanceAsr": "concordance_asr",
     "PhyloLogistic": "phylo_logistic",
+    "PhyloAnova": "phylo_anova",
     "PhyloImpute": "phylo_impute",
     "TraitCorrelation": "trait_correlation",
     "TraitRateMap": "trait_rate_map",