Add subclades to measurement panel outputs

Adds support for filtering and grouping by new subclade annotations.
nextstrain · Aug 29, 2023 · fe6b6f0 · fe6b6f0
1 parent d160e5c
commit fe6b6f0
Show file tree

Hide file tree

Showing 2 changed files with 27 additions and 2 deletions.
diff --git a/scripts/get_antigenic_distances_between_strains.py b/scripts/get_antigenic_distances_between_strains.py
@@ -12,6 +12,7 @@
     parser.add_argument("--titers", required=True, help="raw titers table with information about the source of each titer")
     parser.add_argument("--tree", required=True, help="tree used to identify the given clades")
     parser.add_argument("--clades", required=True, help="clade annotations in a node data JSON")
+    parser.add_argument("--subclades", required=True, help="subclade annotations in a node data JSON")
     parser.add_argument("--haplotypes", required=True, help="haplotype annotations in a node data JSON")
     parser.add_argument("--branch-lengths", required=True, help="branch length annotations including `numdate` calculated by TreeTime")
     parser.add_argument("--frequencies", required=True, help="tip frequencies JSON from augur frequencies")
@@ -114,6 +115,7 @@
     # Load node data (e.g., clade and haplotype).
     node_data = read_node_data([
         args.clades,
+        args.subclades,
         args.haplotypes,
     ])
 
@@ -138,6 +140,7 @@
         {
             "strain": strain,
             "clade": strain_data["clade_membership"],
+            "subclade": strain_data["subclade"],
             "haplotype": strain_data["haplotype"],
             "clade_frequency": frequency_by_clade[strain_data["clade_membership"]],
         }

diff --git a/workflow/snakemake_rules/titer_models.smk b/workflow/snakemake_rules/titer_models.smk
@@ -78,6 +78,7 @@ rule antigenic_distances_between_strains:
     input:
         tree="builds/{build_name}/{segment}/tree.nwk",
         clades="builds/{build_name}/{segment}/clades.json",
+        subclades="builds/{build_name}/{segment}/subclades.json",
         haplotypes="builds/{build_name}/{segment}/haplotypes.json",
         titer_model="builds/{build_name}/{segment}/titers-sub-model/{titer_collection}.json",
         titers="builds/{build_name}/titers/{titer_collection}.tsv",
@@ -97,6 +98,7 @@ rule antigenic_distances_between_strains:
         python3 scripts/get_antigenic_distances_between_strains.py \
             --tree {input.tree} \
             --clades {input.clades} \
+            --subclades {input.subclades} \
             --haplotypes {input.haplotypes} \
             --titer-model {input.titer_model} \
             --titers {input.titers} \
@@ -114,7 +116,15 @@ rule generate_collection_config_json:
         config_json="builds/{build_name}/{segment}/measurements_collection_config/{titer_collection}.json",
     conda: "../envs/nextstrain.yaml"
     params:
-        groupings=["reference_strain", "reference_strain_source", "clade_reference", "haplotype_reference", "source", "serum"],
+        groupings=[
+            "reference_strain",
+            "reference_strain_source",
+            "clade_reference",
+            "subclade_reference",
+            "haplotype_reference",
+            "source",
+            "serum"
+        ],
         fields=[
             "strain",
             "reference_strain",
@@ -127,6 +137,8 @@ rule generate_collection_config_json:
             "reference_date",
             "clade_test",
             "clade_reference",
+            "subclade_test",
+            "subclade_reference",
             "haplotype_test",
             "haplotype_reference",
         ],
@@ -164,7 +176,15 @@ rule export_measurements:
         title=get_titer_collection_title,
         x_axis_label="normalized log2 titer",
         thresholds=[0.0, 2.0],
-        filters=["reference_strain", "reference_strain_source", "clade_reference", "haplotype_reference", "source", "serum"],
+        filters=[
+            "reference_strain",
+            "reference_strain_source",
+            "clade_reference",
+            "subclade_reference",
+            "haplotype_reference",
+            "source",
+            "serum"
+        ],
         include_columns=[
             "test_strain",
             "reference_strain",
@@ -177,6 +197,8 @@ rule export_measurements:
             "reference_date",
             "clade_test",
             "clade_reference",
+            "subclade_test",
+            "subclade_reference",
             "haplotype_test",
             "haplotype_reference",
         ],