phykit 2.1.80__tar.gz → 2.1.82__tar.gz

{phykit-2.1.80 → phykit-2.1.82}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: phykit
-Version: 2.1.80
+Version: 2.1.82
 Home-page: https://github.com/jlsteenwyk/phykit
 Author: Jacob L. Steenwyk
 Author-email: jlsteenwyk@gmail.com

{phykit-2.1.80 → phykit-2.1.82}/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",
@@ -114,6 +118,9 @@ ALIAS_TO_HANDLER: Dict[str, str] = {
     "casr": "concordance_asr",
     "simmap": "stochastic_character_map",
     "scm": "stochastic_character_map",
+    "simmap_summary": "simmap_summary",
+    "smsummary": "simmap_summary",
+    "describe_simmap": "simmap_summary",
     "contmap": "cont_map",
     "cmap": "cont_map",
     "densitymap": "density_map",

{phykit-2.1.80 → phykit-2.1.82}/phykit/phykit.py

@@ -248,10 +248,14 @@ 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)
                     - stochastic character mapping (SIMMAP) of discrete traits
+                simmap_summary (alias: smsummary; describe_simmap)
+                    - per-branch SIMMAP summary with node posteriors
                 cont_map (alias: contmap; cmap)
                     - continuous trait map (contMap) visualization on a phylogeny
                 trait_rate_map (alias: rate_map; branch_rates)
@@ -2004,6 +2008,87 @@ 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(
@@ -4713,6 +4798,79 @@ class Phykit:
         _add_json_argument(parser)
         _run_service(parser, argv, StochasticCharacterMap)
 
+    @staticmethod
+    def simmap_summary(argv):
+        parser = _new_parser(
+            description=textwrap.dedent(
+                f"""\
+                {help_header}
+
+                Run N stochastic character maps and summarize per-branch
+                dwelling time proportions, expected transitions, and
+                posterior state probabilities at each node.
+
+                This extends stochastic_character_map by providing a
+                detailed per-branch summary analogous to
+                phytools::describe.simmap() in R.
+
+                Aliases:
+                  simmap_summary, smsummary, describe_simmap
+                Command line interfaces:
+                  pk_simmap_summary, pk_smsummary, pk_describe_simmap
+
+                Usage:
+                phykit simmap_summary -t <tree> -d <trait_data> -c <trait>
+                  [-m/--model ER|SYM|ARD] [-n/--nsim <int>]
+                  [--seed <int>] [--plot <file>] [--csv <file>] [--json]
+
+                Options
+                =====================================================
+                -t/--tree                   phylogenetic tree file
+                                            (required)
+
+                -d/--trait_data             tab-delimited trait file
+                                            with header row (required)
+
+                -c/--trait                  column name for the
+                                            discrete character trait
+                                            (required)
+
+                -m/--model                  substitution model: ER
+                                            (equal rates), SYM
+                                            (symmetric), or ARD (all
+                                            rates different). Default: ER
+
+                -n/--nsim                   number of stochastic maps
+                                            to simulate (default: 100)
+
+                --seed                      random seed for
+                                            reproducibility
+
+                --plot                      output plot file showing
+                                            tree with posterior pie
+                                            charts at nodes
+
+                --csv                       output CSV file with
+                                            per-branch dwelling
+                                            proportions and node
+                                            posteriors
+
+                --json                      output results as JSON
+                """
+            ),
+        )
+        parser.add_argument("-t", "--tree", type=str, required=True, help=SUPPRESS, metavar="")
+        parser.add_argument("-d", "--trait_data", type=str, required=True, help=SUPPRESS, metavar="")
+        parser.add_argument("-c", "--trait", type=str, required=True, help=SUPPRESS, metavar="")
+        parser.add_argument("-m", "--model", type=str, required=False, default="ER", help=SUPPRESS, metavar="")
+        parser.add_argument("-n", "--nsim", type=int, required=False, default=100, help=SUPPRESS, metavar="")
+        parser.add_argument("--seed", type=int, required=False, default=None, help=SUPPRESS, metavar="")
+        parser.add_argument("--plot", type=str, required=False, default=None, help=SUPPRESS, metavar="")
+        parser.add_argument("--csv", type=str, required=False, default=None, help=SUPPRESS, metavar="")
+        add_plot_arguments(parser)
+        _add_json_argument(parser)
+        _run_service(parser, argv, SimmapSummary)
+
     @staticmethod
     def cont_map(argv):
         parser = _new_parser(
@@ -8685,6 +8843,10 @@ 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:])
 
@@ -8810,6 +8972,10 @@ def stochastic_character_map(argv=None):
     Phykit.stochastic_character_map(sys.argv[1:])
 
 
+def simmap_summary(argv=None):
+    Phykit.simmap_summary(sys.argv[1:])
+
+
 def cont_map(argv=None):
     Phykit.cont_map(sys.argv[1:])
 

{phykit-2.1.80 → phykit-2.1.82}/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")
@@ -75,6 +76,7 @@ Phylomorphospace = _LazyServiceFactory("phykit.services.tree.phylomorphospace",
 PhylogeneticRegression = _LazyServiceFactory("phykit.services.tree.phylogenetic_regression", "PhylogeneticRegression")
 PhylogeneticGLM = _LazyServiceFactory("phykit.services.tree.phylogenetic_glm", "PhylogeneticGLM")
 PhyloLogistic = _LazyServiceFactory("phykit.services.tree.phylo_logistic", "PhyloLogistic")
+SimmapSummary = _LazyServiceFactory("phykit.services.tree.simmap_summary", "SimmapSummary")
 StochasticCharacterMap = _LazyServiceFactory("phykit.services.tree.stochastic_character_map", "StochasticCharacterMap")
 ContMap = _LazyServiceFactory("phykit.services.tree.cont_map", "ContMap")
 DensityMap = _LazyServiceFactory("phykit.services.tree.density_map", "DensityMap")

{phykit-2.1.80 → phykit-2.1.82}/phykit/services/tree/__init__.py

@@ -40,6 +40,7 @@ _EXPORTS = {
     "RobinsonFouldsDistance": "rf_distance",
     "RootTree": "root_tree",
     "Saturation": "saturation",
+    "SimmapSummary": "simmap_summary",
     "Spr": "spr",
     "SpuriousSequence": "spurious_sequence",
     "TerminalBranchStats": "terminal_branch_stats",
@@ -52,6 +53,7 @@ _EXPORTS = {
     "TreenessOverRCV": "treeness_over_rcv",
     "ConcordanceAsr": "concordance_asr",
     "PhyloLogistic": "phylo_logistic",
+    "PhyloAnova": "phylo_anova",
     "PhyloImpute": "phylo_impute",
     "TraitCorrelation": "trait_correlation",
     "TraitRateMap": "trait_rate_map",