Merge branch 'rel-0.3.3'

FastOMA/__init__.py

@@ -1,3 +1,3 @@
 
 __packagename__ = "FastOMA"
-__version__ = "0.3.2"
+__version__ = "0.3.3"

FastOMA/_hog_class.py

@@ -366,7 +366,7 @@ def _sorter_key(sh):
 """
 
 
-def split_hog(hog:HOG, *partitions):
+def split_hog(hog:HOG, level_name:str, *partitions):
     """splits a hog into parts based to the partitions provided.
 
     The partitions need to be a lists of Representatives (or simply members)
@@ -402,4 +402,17 @@ def split_hog(hog:HOG, *partitions):
             h = HOG(subhogs, taxnomic_range=hog.taxlevel, rhogid=hog.rhogid, msa=hog.get_msa(), representatives=partitions[p])
             hogs.append(h)
         return hogs
+    else:
+        rep_to_subhog_list = {rep: hog.get_subhog_path(rep.get_id(), max_depth=-1) for part in partitions for rep in part}
+        for depth in range(max(len(sh) for sh in rep_to_subhog_list.values())):
+            rep_sets = [set(rep_to_subhog_list[rep][depth] for rep in part if len(rep_to_subhog_list[rep]) > depth) for part in partitions]
+            if all(s1.isdisjoint(s2) for s1, s2 in itertools.combinations(rep_sets, 2)):
+                break
+        depth -= 1
+        rep_sets = [set(rep_to_subhog_list[rep][depth] for rep in part if len(rep_to_subhog_list[rep]) > depth) for part in partitions]
+        merged_in = set.union(*(s1.intersection(s2) for s1, s2 in itertools.combinations(rep_sets, 2)))
+        involved_reps = [(r, h[depth].hogid, h[depth].taxname) for r, h in rep_to_subhog_list.items() if len(h)>depth and h[depth] in merged_in]
+        logger.warning(f"{hog} should be split in {len(partitions)} partitions at {level_name}. Not implemented yet.")
+        for rep in involved_reps:
+            logger.warning(f"  - Rep {rep[0]} merged into {rep[1]} at level {rep[2]}")
     #raise RuntimeError("this part of the code needs more thinking")

FastOMA/_infer_subhog.py

@@ -1,7 +1,6 @@
 import collections
 import csv
 import itertools
-from copy import deepcopy
 from pathlib import Path
 import gzip
 
@@ -436,7 +435,7 @@ def infer_genetree_from_msa(self, msa):
         return genetree
 
     def infer_rooted_genetree(self, gene_tree: TreeNode):
-        genetree = deepcopy(gene_tree)
+        genetree = gene_tree.copy()
         if self.conf.gene_rooting_method == "midpoint":
             r_outgroup = genetree.get_midpoint_outgroup()
             genetree.set_outgroup(r_outgroup)  # print("Midpoint rooting is done for gene tree.")
@@ -588,7 +587,7 @@ def merge_subhogs(self, reconciled_genetree:TreeNode, msa:MultipleSeqAlignment):
                 if len(subtrees) > 1:
                     logger.info(f"Representaives of {hogid} are split among {len(subtrees)} candidate subtrees.")
                     split_parts = [list(n.name for n in sub.iter_leaves() if n.hogid == hogid) for sub in subtrees]
-                    split_hogs = split_hog(self.subhogs[hogid], *split_parts)
+                    split_hogs = split_hog(self.subhogs[hogid], self.node_species_tree.name, *split_parts)
                     if split_hogs and len(split_hogs) > 1:
                         # we could split the current hog.
                         self.subhogs.pop(hogid)

nextflow.config

@@ -8,12 +8,12 @@ manifest {
   nextflowVersion = ">=22.10.4"
   defaultBranch = "main"
   doi = "10.1101/2024.01.29.577392"
-  version = "0.3.2"
+  version = "0.3.3"
 }
 
 params {
   container_name = "dessimozlab/fastoma"
-  container_version = "0.3.2"
+  container_version = "0.3.3"
   omamer_db = "https://omabrowser.org/All/LUCA.h5"
   debug_enabled = false
   help = false