MRG: replace lca_utils.LineagePair with tax_utils.LineagePair (#2441

)

Note: PR into #2437

Wanted to try out the new code for myself!

Co-authored-by: N. Tessa Pierce-Ward <ntpierce@gmail.com>
Co-authored-by: Tessa Pierce Ward <bluegenes@users.noreply.github.com>

src/sourmash/lca/__init__.py

@@ -2,7 +2,7 @@
 
 from .lca_db import LCA_Database
 from .lca_utils import (taxlist, zip_lineage, build_tree, find_lca,
-                        gather_assignments, LineagePair, display_lineage,
+                        gather_assignments, display_lineage,
                         count_lca_for_assignments)
 
 from .command_index import index

src/sourmash/lca/command_index.py

@@ -11,7 +11,6 @@
 from sourmash.sourmash_args import load_file_as_signatures
 from sourmash.logging import notify, error, debug, set_quiet
 from . import lca_utils
-from .lca_utils import LineagePair
 from .lca_db import LCA_Database
 from sourmash.sourmash_args import DEFAULT_LOAD_K
 
@@ -26,6 +25,7 @@ def load_taxonomy_assignments(filename, *, delimiter=',', start_column=2,
     The 'assignments' dictionary that's returned maps identifiers to
     lineage tuples.
     """
+    from sourmash.tax.tax_utils import LineagePair
     # parse spreadsheet!
     # CTB note: can't easily switch to FileInputCSV, because of
     # janky way we do/don't handle headers here. See issue #2198.

src/sourmash/lca/lca_db.py

@@ -262,7 +262,8 @@ def load(cls, db_name):
 
         Method specific to this class.
         """
-        from .lca_utils import taxlist, LineagePair
+        from .lca_utils import taxlist
+        from sourmash.tax.tax_utils import LineagePair
 
         if not os.path.isfile(db_name):
             raise ValueError(f"'{db_name}' is not a file and cannot be loaded as an LCA database")
@@ -325,7 +326,7 @@ def load(cls, db_name):
             lid_to_lineage = {}
             lineage_to_lid = {}
             for k, v in lid_to_lineage_2.items():
-                v = dict(v)
+                v = dict( ((x[0], x[1]) for x in v) )
                 vv = []
                 for rank in taxlist():
                     name = v.get(rank, '')

src/sourmash/lca/lca_utils.py

@@ -103,17 +103,6 @@ def display_lineage(lineage, include_strain=True, truncate_empty=True):
 null_names = set(['[Blank]', 'na', 'null'])
 
 
-def make_lineage(lineage_str):
-    "Turn a ; or ,-separated set of lineages into a tuple of LineagePair objs."
-    lin = lineage_str.split(';')
-    if len(lin) == 1:
-        lin = lineage.split(',')
-    lin = [ LineagePair(rank, n) for (rank, n) in zip(taxlist(), lin) ]
-    lin = tuple(lin)
-
-    return lin
-
-
 def build_tree(assignments, initial=None):
     """
     Builds a tree of dictionaries from lists of LineagePair objects
@@ -249,6 +238,7 @@ def pop_to_rank(lin, rank):
 
 def make_lineage(lineage):
     "Turn a ; or ,-separated set of lineages into a tuple of LineagePair objs."
+    from sourmash.tax.tax_utils import LineagePair
     lin = lineage.split(';')
     if len(lin) == 1:
         lin = lineage.split(',')

src/sourmash/tax/__main__.py

@@ -460,7 +460,9 @@ def summarize(args):
     name_seen = set()
     for v in tax_assign.values():
         sofar = []
-        for rank, name in v:
+        for vv in v:
+            name = vv.name
+            rank = vv.rank
             if name not in name_seen:
                 rank_counts[rank] += 1
                 name_seen.add(name)

src/sourmash/tax/tax_utils.py

@@ -161,6 +161,7 @@ def _init_from_lineage_tuples(self):
                     new_lineage[rank_idx] = LineagePair(rank=lin_tup.rank, name=lin_tup.name)
                 else:
                     new_lineage[rank_idx] = lin_tup
+
         # build list of filled ranks
         filled_ranks = [a.rank for a in new_lineage if a.name]
         # set lineage and filled_ranks
@@ -293,7 +294,7 @@ def pop_to_rank(self, rank):
         return new
 
     def lineage_at_rank(self, rank):
-        # non-destructive pop_to_rank. Returns tuple of LineagePairs
+        "non-destructive pop_to_rank. Returns tuple of LineagePairs"
         "Returns tuple of LineagePairs at given rank."
         # are we already above rank?
         if not self.rank_is_filled(rank):