Allow build-specific clade_recency

Set up a build-specific override for colors: clade_recency parameter. Make the default to be empty string "" and have this keep all clades colored. Override for the Nextstrain GISAID profile to specify clade_recency of
- `1` month for "1m" build
- `2` months for "2m" build
- `6` months for "6m" build
- `""` (all months) for "all-time" build

defaults/parameters.yaml

@@ -134,9 +134,9 @@ refine:
 ancestral:
   inference: "joint"
 
-# Gray out clades that are older than 18 months when constructing a color ramp
 colors:
-  clade_recency: 18
+  # Months back to color clades, if "" then all clades are colored
+  clade_recency: ""
 
 # Frequencies settings
 frequencies:

docs/src/reference/workflow-config-file.rst

@@ -936,6 +936,33 @@ no_timetree
 -  description: Do not produce a time tree.
 -  default: ``false``
 
+colors
+------
+
+-  type: object
+-  description: Parameters for assigning colors in ``scripts/assign-colors.py``
+-  examples:
+
+.. code:: yaml
+
+   colors:
+     default:
+       clade_recency: ""
+     global-6m:
+       # Override clade recency colors for "global-6m" build
+       clade_recency: 6
+
+Each named traits configuration (``default`` or build-named) supports the following attributes:
+
+.. contents::
+   :local:
+
+clade_recency
+~~~~~~~~~~~~~
+
+-  type: integer
+-  description: if provided, restrict to clades found in tree within X months of present
+-  default: ````
 
 traits
 ------

nextstrain_profiles/nextstrain-gisaid/builds.yaml

@@ -874,6 +874,68 @@ subsampling:
       max_sequences: 75
       exclude: "--exclude-where 'region!=Oceania'"
 
+# if different traits should be reconstructed for some builds, specify here
+# otherwise the default trait config in defaults/parameters.yaml will used
+colors:
+  reference:
+    clade_recency: ""
+  global_1m:
+    clade_recency: 1
+  global_2m:
+    clade_recency: 2
+  global_6m:
+    clade_recency: 6
+  global_all-time:
+    clade_recency: ""
+  africa_1m:
+    clade_recency: 1
+  africa_2m:
+    clade_recency: 2
+  africa_6m:
+    clade_recency: 6
+  africa_all-time:
+    clade_recency: ""
+  asia_1m:
+    clade_recency: 1
+  asia_2m:
+    clade_recency: 2
+  asia_6m:
+    clade_recency: 6
+  asia_all-time:
+    clade_recency: ""
+  europe_1m:
+    clade_recency: 1
+  europe_2m:
+    clade_recency: 2
+  europe_6m:
+    clade_recency: 6
+  europe_all-time:
+    clade_recency: ""
+  north-america_1m:
+    clade_recency: 1
+  north-america_2m:
+    clade_recency: 2
+  north-america_6m:
+    clade_recency: 6
+  north-america_all-time:
+    clade_recency: ""
+  oceania_1m:
+    clade_recency: 1
+  oceania_2m:
+    clade_recency: 2
+  oceania_6m:
+    clade_recency: 6
+  oceania_all-time:
+    clade_recency: ""
+  south-america_1m:
+    clade_recency: 1
+  south-america_2m:
+    clade_recency: 2
+  south-america_6m:
+    clade_recency: 6
+  south-america_all-time:
+    clade_recency: ""
+
 # if different traits should be reconstructed for some builds, specify here
 # otherwise the default trait config in defaults/parameters.yaml will used
 traits:

scripts/assign-colors.py

@@ -25,7 +25,7 @@ def date_within_last_n_months(date_str, cutoff_date):
     parser.add_argument('--color-schemes', type=str, required=True, help="input color schemes file")
     parser.add_argument('--metadata', type=str, help="if provided, restrict colors to only those found in metadata")
     parser.add_argument('--clade-node-data', type=str, help="if provided, restrict to only those clades found in tree")
-    parser.add_argument('--clade-recency', type=int, help="if provided, restrict to clades found in tree within X months of present")
+    parser.add_argument('--clade-recency', type=int, nargs='?', const=None, help="if provided, restrict to clades found in tree within X months of present")
     parser.add_argument('--output', type=str, required=True, help="output colors tsv")
     args = parser.parse_args()
 
@@ -61,7 +61,7 @@ def date_within_last_n_months(date_str, cutoff_date):
             import json
             clades = json.load(fh)['nodes']
 
-        if args.clade_recency and args.metadata:
+        if args.clade_recency is not None and args.metadata:
             # Calculate the cutoff date based on clade_recency (number of months ago from today)
             cutoff_date = datetime.today() - timedelta(days=args.clade_recency * 30)  # approximate months as 30 days
 

workflow/snakemake_rules/common.smk

@@ -168,6 +168,12 @@ def _get_metadata_by_wildcards(wildcards):
     """
     return _get_metadata_by_build_name(wildcards.build_name)
 
+def _get_clade_recency_for_wildcards(wildcards):
+    if wildcards.build_name in config["colors"] and 'clade_recency' in config["colors"][wildcards.build_name]:
+        return config["colors"][wildcards.build_name]["clade_recency"]
+    else:
+        return config["colors"]["default"]["clade_recency"]
+
 def _get_trait_columns_by_wildcards(wildcards):
     if wildcards.build_name in config["traits"]:
         return config["traits"][wildcards.build_name]["columns"]

workflow/snakemake_rules/main_workflow.smk

@@ -1102,7 +1102,7 @@ rule colors:
         metadata="results/{build_name}/metadata_adjusted.tsv.xz",
         clades = rules.clades.output.clade_data
     params:
-        clade_recency = config["colors"]["clade_recency"]
+        clade_recency = _get_clade_recency_for_wildcards
     output:
         colors = "results/{build_name}/colors.tsv"
     log: