MRG: add utilities for using ictv taxonomic ranks with sourmash tax (…

…#2608)

doc/command-line.md

@@ -551,8 +551,8 @@ The sourmash `tax` or `taxonomy` commands integrate taxonomic
  `gather` command (we cannot combine separate `gather` runs for the
  same query). For supported databases (e.g. GTDB, NCBI), we provide
  taxonomy csv files, but they can also be generated for user-generated
- databases. As of v4.8, some sourmash taxonomy commands can also use `LIN`
- lineage information. For more information, see [databases](databases.md).
+ databases. As of v4.8 and 4.8.6, respectively, some sourmash taxonomy
+ commands can also use `LIN` or `ICTV` lineage information.
 
 `tax` commands rely upon the fact that `gather` provides both the total
  fraction of the query matched to each database matched, as well as a

src/sourmash/cli/tax/annotate.py

@@ -78,6 +78,13 @@ def subparser(subparsers):
         default=False,
         help="use LIN taxonomy in place of standard taxonomic ranks.  Note that the taxonomy CSV must contain LIN lineage information.",
     )
+    subparser.add_argument(
+        "--ictv",
+        "--ictv-taxonomy",
+        action="store_true",
+        default=False,
+        help="use ICTV taxonomy in place of standard taxonomic ranks.  Note that the taxonomy CSV must contain ICTV ranks.",
+    )
 
 
 def main(args):

src/sourmash/cli/tax/genome.py

@@ -124,6 +124,13 @@ def subparser(subparsers):
         default=None,
         help="CSV containing 'name', 'lin' columns, where 'lin' is the lingroup prefix. Will restrict classification to these groups.",
     )
+    subparser.add_argument(
+        "--ictv",
+        "--ictv-taxonomy",
+        action="store_true",
+        default=False,
+        help="use ICTV taxonomy in place of standard taxonomic ranks.  Note that the taxonomy CSV must contain ICTV ranks.",
+    )
     add_tax_threshold_arg(subparser, 0.1)
     add_rank_arg(subparser)
 

src/sourmash/cli/tax/metagenome.py

@@ -116,6 +116,13 @@ def subparser(subparsers):
         default=None,
         help="CSV containing 'name', 'lin' columns, where 'lin' is the lingroup prefix. Will produce a 'lingroup' report containing taxonomic summarization for each group.",
     )
+    subparser.add_argument(
+        "--ictv",
+        "--ictv-taxonomy",
+        action="store_true",
+        default=False,
+        help="use ICTV taxonomy in place of standard taxonomic ranks.  Note that the taxonomy CSV must contain ICTV ranks.",
+    )
     add_rank_arg(subparser)
 
 

src/sourmash/cli/tax/summarize.py

@@ -57,6 +57,13 @@ def subparser(subparsers):
         default=False,
         help="use LIN taxonomy in place of standard taxonomic ranks.",
     )
+    subparser.add_argument(
+        "--ictv",
+        "--ictv-taxonomy",
+        action="store_true",
+        default=False,
+        help="use ICTV taxonomy in place of standard taxonomic ranks.  Note that the taxonomy CSV must contain ICTV ranks.",
+    )
 
 
 def main(args):

src/sourmash/tax/__main__.py

@@ -18,6 +18,7 @@
     RankLineageInfo,
     LINLineageInfo,
     AnnotateTaxResult,
+    ICTVRankLineageInfo,
 )
 
 usage = """
@@ -82,6 +83,7 @@ def metagenome(args):
             keep_identifier_versions=args.keep_identifier_versions,
             force=args.force,
             lins=args.lins,
+            ictv=args.ictv,
         )
         available_ranks = tax_assign.available_ranks
     except ValueError as exc:
@@ -113,6 +115,7 @@ def metagenome(args):
             keep_full_identifiers=args.keep_full_identifiers,
             keep_identifier_versions=args.keep_identifier_versions,
             lins=args.lins,
+            ictv=args.ictv,
         )
     except ValueError as exc:
         error(f"ERROR: {str(exc)}")
@@ -258,6 +261,7 @@ def genome(args):
             keep_identifier_versions=args.keep_identifier_versions,
             force=args.force,
             lins=args.lins,
+            ictv=args.ictv,
         )
         available_ranks = tax_assign.available_ranks
 
@@ -297,6 +301,7 @@ def genome(args):
             keep_full_identifiers=args.keep_full_identifiers,
             keep_identifier_versions=args.keep_identifier_versions,
             lins=args.lins,
+            ictv=args.ictv,
         )
 
     except ValueError as exc:
@@ -402,6 +407,7 @@ def annotate(args):
             keep_identifier_versions=args.keep_identifier_versions,
             force=args.force,
             lins=args.lins,
+            ictv=args.ictv,
         )
 
     except ValueError as exc:
@@ -466,6 +472,7 @@ def annotate(args):
                             raw=row,
                             id_col=id_col,
                             lins=args.lins,
+                            ictv=args.ictv,
                             keep_full_identifiers=args.keep_full_identifiers,
                             keep_identifier_versions=args.keep_identifier_versions,
                         )
@@ -591,6 +598,7 @@ def summarize(args):
             keep_full_identifiers=args.keep_full_identifiers,
             keep_identifier_versions=args.keep_identifier_versions,
             lins=args.lins,
+            ictv=args.ictv,
         )
     except ValueError as exc:
         error("ERROR while loading taxonomies!")
@@ -637,6 +645,8 @@ def summarize(args):
                 rank = lineage[-1].rank
                 if args.lins:
                     inf = LINLineageInfo(lineage=lineage)
+                elif args.ictv:
+                    inf = ICTVRankLineageInfo(lineage=lineage)
                 else:
                     inf = RankLineageInfo(lineage=lineage)
                 lin = inf.display_lineage()