Remove default sp_naming_function in PhyloTree.

The one we had makes no sense for modern trees. Better
not to set any function unless the user chooses it.

ete4/nexml/_nexml_tree.py

@@ -1,7 +1,6 @@
 import sys
 from ._nexml import MixedContainer, FloatTree, TreeFloatEdge, TreeNode, LiteralMeta
 from .. import PhyloTree
-from ..phylo.phylotree import _parse_species
 from ..parser.newick import loads
 
 class Children(list):
@@ -58,7 +57,7 @@ def _set_children(self, value):
     name = property(fget=_get_name, fset=_set_name)
 
     def __init__(self, newick=None, alignment=None, alg_format="fasta", \
-                 sp_naming_function=_parse_species, parser=0):
+                 sp_naming_function=None, parser=0):
 
         self.nexml_tree = FloatTree()
         self.nexml_tree.set_anyAttributes_({'xsi:type': 'FloatTree'})

ete4/phylo/phylotree.py

@@ -15,8 +15,6 @@
 
 __all__ = ["PhyloTree"]
 
-def _parse_species(name):
-    return name[:3] if name is not None else ''
 
 def is_dup(n):
     return n.props.get("evoltype") == "D"
@@ -272,7 +270,7 @@ class PhyloTree(Tree):
     """
 
     def __init__(self, newick=None, children=None, alignment=None,
-                 alg_format="fasta", sp_naming_function=_parse_species,
+                 alg_format="fasta", sp_naming_function=None,
                  parser=None):
         """
         :param newick: If not None, initializes the tree from a newick,

tests/test_phylotree.py

@@ -54,7 +54,8 @@ def test_link_alignmets(self):
         alg1 = SeqGroup(fasta)
         alg2 = SeqGroup(iphylip, format="iphylip")
 
-        t = PhyloTree("(((seqA,seqB),seqC),seqD);", alignment=fasta, alg_format="fasta")
+        t = PhyloTree("(((seqA,seqB),seqC),seqD);", alignment=fasta, alg_format="fasta",
+                      sp_naming_function=lambda name: name[:3])
 
         for l in t.leaves():
             self.assertEqual(l.props.get('sequence'), alg1.get_seq(l.name))
@@ -69,7 +70,8 @@ def test_get_sp_overlap_on_all_descendants(self):
         """ Tests ortholgy prediction using the sp overlap"""
         # Creates a gene phylogeny with several duplication events at
         # different levels.
-        t = PhyloTree(example_tree)
+        t = PhyloTree(example_tree,
+                      sp_naming_function=lambda name: name[:3])
 
         # Scans the tree using the species overlap algorithm and detect all
         # speciation and duplication events
@@ -158,19 +160,22 @@ def test_get_sp_overlap_on_all_descendants(self):
         self.assertEqual(expected_orthologs, orthologs)
 
         # Test different sos_thr
-        t = PhyloTree('(((SP1_a, SP2_a), (SP3_a, SP1_b)), (SP1_c, SP2_c));')
+        t = PhyloTree('(((SP1_a, SP2_a), (SP3_a, SP1_b)), (SP1_c, SP2_c));',
+                      sp_naming_function=lambda name: name[:3])
         seed = t['SP1_a']
         events = t.get_descendant_evol_events(0.1)
         self.assertEqual(t.common_ancestor([seed, 'SP3_a']).props.get('evoltype'), 'D')
         self.assertEqual(t.common_ancestor([seed, 'SP1_c']).props.get('evoltype'), 'D')
 
-        t = PhyloTree('(((SP1_a, SP2_a), (SP3_a, SP1_b)), (SP1_c, SP2_c));')
+        t = PhyloTree('(((SP1_a, SP2_a), (SP3_a, SP1_b)), (SP1_c, SP2_c));',
+                      sp_naming_function=lambda name: name[:3])
         seed = t['SP1_a']
         events = t.get_descendant_evol_events(0.5)
         self.assertEqual(t.common_ancestor([seed, 'SP3_a']).props.get('evoltype'), 'S')
         self.assertEqual(t.common_ancestor([seed, 'SP1_c']).props.get('evoltype'), 'D')
 
-        t = PhyloTree('(((SP1_a:1, SP2_a:1):1, (SP3_a:1, SP1_b:1):1):1, (SP1_c:1, SP2_c:1):1):0;')
+        t = PhyloTree('(((SP1_a:1, SP2_a:1):1, (SP3_a:1, SP1_b:1):1):1, (SP1_c:1, SP2_c:1):1):0;',
+                      sp_naming_function=lambda name: name[:3])
         seed = t['SP1_a']
         events = seed.get_my_evol_events(0.75)
         self.assertEqual(t.common_ancestor([seed, 'SP3_a']).props.get('evoltype'), 'S')
@@ -180,7 +185,8 @@ def test_get_sp_overlap_on_a_seed(self):
         """ Tests ortholgy prediction using sp overlap"""
         # Creates a gene phylogeny with several duplication events at
         # different levels.
-        t = PhyloTree(example_tree)
+        t = PhyloTree(example_tree,
+                      sp_naming_function=lambda name: name[:3])
 
         # Scans the tree using the species overlap algorithm
         seed = t['Hsa_001']
@@ -261,19 +267,22 @@ def test_get_sp_overlap_on_a_seed(self):
         self.assertEqual(expected_orthologs, orthologs)
 
         # Test different sos_thr
-        t = PhyloTree('(((SP1_a:1, SP2_a:1):1, (SP3_a:1, SP1_b:1):1):1, (SP1_c:1, SP2_c:1):1):0;')
+        t = PhyloTree('(((SP1_a:1, SP2_a:1):1, (SP3_a:1, SP1_b:1):1):1, (SP1_c:1, SP2_c:1):1):0;',
+                      sp_naming_function=lambda name: name[:3])
         seed = t['SP1_a']
         events = seed.get_my_evol_events(0.1)
         self.assertEqual(t.common_ancestor([seed, 'SP3_a']).props.get('evoltype'), 'D')
         self.assertEqual(t.common_ancestor([seed, 'SP1_c']).props.get('evoltype'), 'D')
 
-        t = PhyloTree('(((SP1_a:1, SP2_a:1):1, (SP3_a:1, SP1_b:1):1):1, (SP1_c:1, SP2_c:1):1):0;')
+        t = PhyloTree('(((SP1_a:1, SP2_a:1):1, (SP3_a:1, SP1_b:1):1):1, (SP1_c:1, SP2_c:1):1):0;',
+                      sp_naming_function=lambda name: name[:3])
         seed = t['SP1_a']
         events = seed.get_my_evol_events(0.50)
         self.assertEqual(t.common_ancestor([seed, 'SP3_a']).props.get('evoltype'), 'S')
         self.assertEqual(t.common_ancestor([seed, 'SP1_c']).props.get('evoltype'), 'D')
 
-        t = PhyloTree('(((SP1_a:1, SP2_a:1):1, (SP3_a:1, SP1_b:1):1):1, (SP1_c:1, SP2_c:1):1):0;')
+        t = PhyloTree('(((SP1_a:1, SP2_a:1):1, (SP3_a:1, SP1_b:1):1):1, (SP1_c:1, SP2_c:1):1):0;',
+                      sp_naming_function=lambda name: name[:3])
         seed = t['SP1_a']
         events = seed.get_my_evol_events(0.75)
         self.assertEqual(t.common_ancestor([seed, 'SP3_a']).props.get('evoltype'), 'S')
@@ -284,8 +293,10 @@ def test_reconciliation(self):
         gene_tree_nw = '((Dme_001,Dme_002),(((Cfa_001,Mms_001),((Hsa_001,Ptr_001),Mmu_001)),(Ptr_002,(Hsa_002,Mmu_002))));'
         species_tree_nw = "((((Hsa, Ptr), Mmu), (Mms, Cfa)), Dme);"
 
-        genetree = PhyloTree(gene_tree_nw)
-        sptree = PhyloTree(species_tree_nw)
+        genetree = PhyloTree(gene_tree_nw,
+                             sp_naming_function=lambda name: name[:3])
+        sptree = PhyloTree(species_tree_nw,
+                           sp_naming_function=lambda name: name[:3])
 
         recon_tree, events = genetree.reconcile(sptree)
         # Check that reconcilied tree nodes have the correct lables:
@@ -299,7 +310,8 @@ def test_miscelaneus(self):
         """ Test several things """
         # Creates a gene phylogeny with several duplication events at
         # different levels.
-        t = PhyloTree(example_tree)
+        t = PhyloTree(example_tree,
+                      sp_naming_function=lambda name: name[:3])
 
         # Create a dictionary with relative ages for the species present in
         # the phylogenetic tree.  Note that ages are only relative numbers to
@@ -337,7 +349,8 @@ def test_miscelaneus(self):
         self.assertEqual(set([sp for sp in t.iter_species()]), set(sp2age.keys()))
 
     def test_collapse(self):
-        t = PhyloTree('((Dme_001,Dme_002),(((Cfa_001,Mms_001),((((Hsa_001,Hsa_001),Ptr_001),Mmu_001),((Hsa_004,Ptr_004),Mmu_004))),(Ptr_002,(Hsa_002,Mmu_002))));')
+        t = PhyloTree('((Dme_001,Dme_002),(((Cfa_001,Mms_001),((((Hsa_001,Hsa_001),Ptr_001),Mmu_001),((Hsa_004,Ptr_004),Mmu_004))),(Ptr_002,(Hsa_002,Mmu_002))));',
+                      sp_naming_function=lambda name: name[:3])
         for n in t.traverse():
             n.dist = 1
             n.support = 1

tests/test_tree.py

@@ -1457,7 +1457,8 @@ def test_monophyly(self):
         self.assertEqual(monotype, 'paraphyletic')
 
         # Now with unrooted trees, and using species instead of names.
-        t = PhyloTree('(aaa1, (aaa3, (aaa4, (bbb1, bbb2))));')
+        t = PhyloTree('(aaa1, (aaa3, (aaa4, (bbb1, bbb2))));',
+                      sp_naming_function=lambda name: name[:3])
         # ─┬╴aaa1        # the species will be 'aaa' for this node, etc.
         #  ╰─┬╴aaa3
         #    ╰─┬╴aaa4
@@ -1469,50 +1470,57 @@ def test_monophyly(self):
                          (True, 'monophyletic', set()))
 
         # Variations on that tree.
-        t = PhyloTree('(aaa1, (bbb3, (aaa4, (bbb1, bbb2))));')
+        t = PhyloTree('(aaa1, (bbb3, (aaa4, (bbb1, bbb2))));',
+                      sp_naming_function=lambda name: name[:3])
         is_mono, _, extra = t.check_monophyly(values={'aaa'},
                                               prop='species', unrooted=True)
         self.assertFalse(is_mono)
         self.assertEqual(extra, {t['bbb3']})
 
-        t = PhyloTree('(aaa1, (aaa3, (aaa4, (bbb1, bbb2))));')
+        t = PhyloTree('(aaa1, (aaa3, (aaa4, (bbb1, bbb2))));',
+                      sp_naming_function=lambda name: name[:3])
         is_mono, _, extra = t.check_monophyly(values={'bbb'},
                                               prop='species', unrooted=True)
         self.assertTrue(is_mono)
         self.assertEqual(extra, set())
 
-        t = PhyloTree('(aaa1, (aaa3, (aaa4, (bbb1, ccc2))));')
+        t = PhyloTree('(aaa1, (aaa3, (aaa4, (bbb1, ccc2))));',
+                      sp_naming_function=lambda name: name[:3])
         is_mono, _, extra = t.check_monophyly(values={'bbb', 'ccc'},
                                               prop='species', unrooted=True)
         self.assertTrue(is_mono)
         self.assertEqual(extra, set())
 
-        t = PhyloTree('(aaa1, (aaa3, (bbb4, (bbb1, bbb2))));')
+        t = PhyloTree('(aaa1, (aaa3, (bbb4, (bbb1, bbb2))));',
+                      sp_naming_function=lambda name: name[:3])
         is_mono, _, extra = t.check_monophyly(values={'bbb4', 'bbb2'},
                                               prop='name', unrooted=True)
         self.assertFalse(is_mono)
         self.assertEqual(extra, {t['bbb1']})
 
-        t = PhyloTree('(aaa1, (aaa3, (bbb4, (bbb1, bbb2))));')
+        t = PhyloTree('(aaa1, (aaa3, (bbb4, (bbb1, bbb2))));',
+                      sp_naming_function=lambda name: name[:3])
         is_mono, _, extra = t.check_monophyly(values={'bbb1', 'bbb2'},
                                               prop='name', unrooted=True)
         self.assertTrue(is_mono)
         self.assertEqual(extra, set())
 
-        t = PhyloTree('(aaa1, aaa3, (aaa4, (bbb1, bbb2)));')
+        t = PhyloTree('(aaa1, aaa3, (aaa4, (bbb1, bbb2)));',
+                      sp_naming_function=lambda name: name[:3])
         is_mono, _, extra = t.check_monophyly(values={'aaa'},
                                               prop='species', unrooted=True)
         self.assertTrue(is_mono)
         self.assertEqual(extra, set())
 
-        t = PhyloTree('(aaa1, bbb3, (aaa4, (bbb1, bbb2)));')
+        t = PhyloTree('(aaa1, bbb3, (aaa4, (bbb1, bbb2)));',
+                      sp_naming_function=lambda name: name[:3])
         is_mono, _, extra = t.check_monophyly(values={'aaa'},
                                               prop='species', unrooted=True)
         self.assertFalse(is_mono)
         self.assertEqual(extra, {t['bbb3']})
 
         # # Check monophyly randomization test
-        # t = PhyloTree()
+        # t = PhyloTree(,
         # t.populate(100)
         # ancestor = t.common_ancestor(['aaaaaaaaaa', 'aaaaaaaaab', 'aaaaaaaaac'])
         # all_nodes = list(t.descendants())

tests/test_treematcher.py

@@ -67,7 +67,7 @@ def test_search():
 
 
 def test_safer():
-    t = PhyloTree('(a,(b,c));')
+    t = PhyloTree('(a,(b,c));', sp_naming_function=lambda name: name)
 
     tp_unsafe = tm.TreePattern('("node.get_species()=={\'c\'}",'
                                '  node.species=="b")')