Process data in ingest to add url field to metadata [#25]

This allows Auspice to automatically link the accession number back to
Genbank in the detailed node popover.

ingest/defaults/config.yaml

@@ -128,6 +128,7 @@ curate:
     - full_authors
     - authors
     - institution
+    - url
 nextclade:
   dataset_name: "nextstrain/yellow-fever/prM-E"
   field_map: "defaults/nextclade_field_map.tsv"

ingest/rules/curate.smk

@@ -59,7 +59,7 @@ rule curate:
         all_geolocation_rules="data/all-geolocation-rules.tsv",
         annotations=config["curate"]["annotations"],
     output:
-        metadata="data/all_metadata.tsv",
+        metadata=temp("data/all_metadata_intermediate.tsv"),
         sequences="results/sequences.fasta",
     log:
         "logs/curate.txt",
@@ -116,6 +116,23 @@ rule curate:
         """
 
 
+rule add_genbank_url:
+    input:
+        metadata=temp("data/all_metadata_intermediate.tsv"),
+    output:
+        metadata="data/all_metadata.tsv",
+    log:
+        "logs/add_genbank_url",
+    benchmark:
+        "benchmarks/add_genbank_url",
+    shell:
+        r"""
+        csvtk mutate2 -tl \
+          -n url \
+          -e '"https://www.ncbi.nlm.nih.gov/nuccore/" + $accession' \
+          {input.metadata:q} > {output.metadata:q} 2> {log:q}
+        """
+
 rule subset_metadata:
     input:
         metadata="data/all_metadata.tsv",