Improve how 'subtype_query' is defined

Brings the config structure in line with all of the other config values,
but this necessitates a helper function as this config value is accessed
before all the wildcards are set (because it's so early in the
pipeline).

This also greatly improves the error handling if we forget to add the
appropriate subtype to the config, or leave out 'subtype_query'
altogether.

Inspired by <#104 (comment)>

genome-focused/config.yaml

@@ -45,12 +45,13 @@ dropped_strains: config/{subtype}/dropped_strains_{subtype}.txt
 clades_file: clade-labeling/h5n1-clades.tsv # use H5N1 clades
 description: config/{subtype}/description.md
 
-## Subtype query - this structure is different from all other config parameters, requiring
-#  a key for each of the subtypes defined above in 'builds'. The string will be supplied to
-#  augur filter's --query argument.
+## Subtype query - note that you cannot vary this based on segment/time wildcards
+#  as the filtering to subtype is independent (upstream) of those wildcards.
+#  Note also that you aren't limited to the "subtype" metadata field - any valid
+#  `augur filter --query` argument is ok.
 subtype_query:
-    "h5n1-cattle-outbreak": "genoflu in 'B3.13'"
-    "h5n1-d1.1": "genoflu in 'D1.1'"
+    "h5n1-cattle-outbreak/*/*": "genoflu in 'B3.13'"
+    "h5n1-d1.1/*/*": "genoflu in 'D1.1'"
 
 #### Rule-specific parameters ####
 # The formatting here represents the three-tiered nature of the avian-flu build which

rules/config.smk

@@ -247,3 +247,16 @@ def expand_target_patterns():
                             targets.append(target)
 
     return targets
+
+def resolve_subtype_query():
+    """
+    Returns a Snakemake Input function to resolve the `subtype_query` config parameter.
+
+    NOTE: This is subtly different from `resolve_config_value` as this config value is used
+    in the snakemake pipeline before we have {segment} or {time} wildcards set up, or if we do
+    have (some of) these wildcards set up then we explicitly do not use them. In other words,
+    the subtype query value does not vary across segment or time wildcards.
+    """
+    def resolve(wildcards):
+        return resolve_config_value('subtype_query')({"subtype": wildcards.subtype, 'segment': '*', 'time': '*'})
+    return resolve

rules/main.smk

@@ -183,7 +183,7 @@ rule filter_sequences_by_subtype:
     output:
         sequences = "results/{subtype}/{segment}/sequences.fasta",
     params:
-        subtypes=lambda w: config['subtype_query'][w.subtype],
+        subtypes = resolve_subtype_query(),
     shell:
         """
         augur filter \
@@ -199,7 +199,7 @@ rule filter_metadata_by_subtype:
     output:
         metadata = "results/{subtype}/metadata.tsv",
     params:
-        subtypes= lambda w: config['subtype_query'][w.subtype],
+        subtypes= resolve_subtype_query(),
     shell:
         """
         augur filter \

segment-focused/config.yaml

@@ -51,14 +51,16 @@ dropped_strains: config/{subtype}/dropped_strains_{subtype}.txt
 clades_file: clade-labeling/{subtype}-clades.tsv
 description: config/description_gisaid.md
 
-## Subtype query - this structure is different from all other config parameters, requiring
-#  a key for each of the subtypes defined above in 'builds'. The string will be supplied to
-#  augur filter's --query argument.
+## Subtype query - note that you cannot vary this based on segment/time wildcards
+#  as the filtering to subtype is independent (upstream) of those wildcards.
+#  Note also that you aren't limited to the "subtype" metadata field - any valid
+#  `augur filter --query` argument is ok.
 subtype_query:
-    "h5nx": "subtype in ['h5n1', 'h5n2', 'h5n3', 'h5n4', 'h5n5', 'h5n6', 'h5n7', 'h5n8', 'h5n9']"
-    "h5n1": "subtype in ['h5n1']"
-    "h7n9": "subtype in ['h7n9']"
-    "h9n2": "subtype in ['h9n2']"
+    "h5nx/*/*": "subtype in ['h5n1', 'h5n2', 'h5n3', 'h5n4', 'h5n5', 'h5n6', 'h5n7', 'h5n8', 'h5n9']"
+    "h5n1/*/*": "subtype in ['h5n1']"
+    "h7n9/*/*": "subtype in ['h7n9']"
+    "h9n2/*/*": "subtype in ['h9n2']"
+
 
 #### Rule-specific parameters ####
 # The formatting here represents the three-tiered nature of the avian-flu build which