Merge branch 'fix-completeness-score' into dev

FastOMA/_hog_class.py

@@ -90,7 +90,6 @@ def __init__(self, input_instantiate, taxnomic_range:TreeNode, rhogid:str, msa:O
             self._msa = MultipleSeqAlignment([only_protein])
             self._subhogs = []
             self._dubious_members = set()
-            self._num_species_tax_speciestree = 1  # at the leaf level, there is only one species in the species tree
             # <<class 'Bio.Align.MultipleSeqAlignment'> instance (1 records of length 314) at 7f0e86c713d0>
 
         elif msa and all(isinstance(x, HOG) for x in input_instantiate):
@@ -102,7 +101,6 @@ def __init__(self, input_instantiate, taxnomic_range:TreeNode, rhogid:str, msa:O
                 hog_members |= sub_hog.get_members()
             self._members = hog_members  # set.union(*tup)    # this also include dubious_members
             self._subhogs = list(input_instantiate)  # full members of subhog, children
-            self._num_species_tax_speciestree = taxnomic_range.size
             self._representatives = [r for r in representatives]
             representatives_id = set(r.get_id() for r in representatives)
             assert len(representatives_id) == len(self._representatives), "Non-unique ids among representatives"
@@ -337,7 +335,7 @@ def _sorter_key(sh):
             #hog_elemnt = ET.Element('orthologGroup', attrib={"id": str(self._hogid)})
             hog_elemnt = ET.Element('orthologGroup', attrib={"id": str(self._hogid)}, )
             num_species_tax_hog = len(set([i.split("||")[1] for i in self._members]))  #  'tr|H2MU14|H2MU14_ORYLA||ORYLA||1056022282'
-            completeness_score = round(num_species_tax_hog/self._num_species_tax_speciestree, 4)
+            completeness_score = round(num_species_tax_hog/self._tax_now.size, 4)
             property_element = ET.SubElement(hog_elemnt, "score", attrib={"id": "CompletenessScore", "value": str(completeness_score)})
             property_element = ET.SubElement(hog_elemnt, "property", attrib={"name": "TaxRange", "value": str(self._tax_now.name)})
 

FastOMA/_utils_subhog.py

@@ -276,11 +276,9 @@ def prepare_species_tree(rhog_i: List[SeqRecord], species_tree: Tree, rhogid: st
     for n in species_tree.traverse():
         n.add_feature("size", len(n))
 
-    first_common_ancestor_name = species_tree.get_common_ancestor(species_names_uniqe).name
-    species_tree.prune(species_names_uniqe, preserve_branch_length=True)
-    # todo check internal node with one child, we need to report for i or not ?
-    species_tree.name = first_common_ancestor_name
-    return species_tree, species_names_rhog, prot_names_rhog
+    mrca = species_tree.get_common_ancestor(species_names_uniqe)
+    mrca.prune(species_names_uniqe, preserve_branch_length=True)
+    return mrca, species_names_rhog, prot_names_rhog
 
 
 def label_sd_internal_nodes(tree_out, threshold_dubious_sd):