Merge pull request #79 from nextstrain/remove-ncov-ingest-geolocation…

…-rules

ingest: Remove use of ncov-ingest geolocation rules

ingest/defaults/config.yaml

@@ -34,10 +34,6 @@ ncbi_datasets_fields:
 
 # Config parameters related to the curate pipeline
 curate:
-  # URL pointed to public generalized geolocation rules
-  # For the Nextstrain team, this is currently
-  # 'https://raw.githubusercontent.com/nextstrain/ncov-ingest/master/source-data/gisaid_geoLocationRules.tsv'
-  geolocation_rules_url: 'https://raw.githubusercontent.com/nextstrain/ncov-ingest/master/source-data/gisaid_geoLocationRules.tsv'
   # The path to the local geolocation rules within the pathogen repo
   # The path should be relative to the ingest directory.
   local_geolocation_rules: 'defaults/geolocation-rules.tsv'

ingest/rules/curate.smk

@@ -13,29 +13,6 @@ Parameters are expected to be defined in `config.curate`.
 """
 
 
-rule fetch_general_geolocation_rules:
-    output:
-        general_geolocation_rules="data/general-geolocation-rules.tsv",
-    params:
-        geolocation_rules_url=config["curate"]["geolocation_rules_url"],
-    shell:
-        """
-        curl {params.geolocation_rules_url} > {output.general_geolocation_rules}
-        """
-
-
-rule concat_geolocation_rules:
-    input:
-        general_geolocation_rules="data/general-geolocation-rules.tsv",
-        local_geolocation_rules=config["curate"]["local_geolocation_rules"],
-    output:
-        all_geolocation_rules="data/all-geolocation-rules.tsv",
-    shell:
-        """
-        cat {input.general_geolocation_rules} {input.local_geolocation_rules} >> {output.all_geolocation_rules}
-        """
-
-
 def format_field_map(field_map: dict[str, str]) -> str:
     """
     Format dict to `"key1"="value1" "key2"="value2"...` for use in shell commands.
@@ -46,7 +23,7 @@ def format_field_map(field_map: dict[str, str]) -> str:
 rule curate:
     input:
         sequences_ndjson="data/sequences.ndjson",
-        all_geolocation_rules="data/all-geolocation-rules.tsv",
+        geolocation_rules=config["curate"]["local_geolocation_rules"],
         annotations=config["curate"]["annotations"],
     output:
         metadata="data/all_metadata.tsv",
@@ -90,7 +67,7 @@ rule curate:
                 --authors-field {params.authors_field} \
                 --default-value {params.authors_default_value} \
             | augur curate apply-geolocation-rules \
-                --geolocation-rules {input.all_geolocation_rules} \
+                --geolocation-rules {input.geolocation_rules} \
             | ./scripts/fix-zika-strain-names.py \
             | augur curate apply-record-annotations \
                 --annotations {input.annotations} \