Merge pull request #40 from nextstrain/fix-phylo-input

Fix phylo input
nextstrain · Apr 15, 2024 · c81b01e · c81b01e
2 parents 416c22d + 9b9d1b2
commit c81b01e
Show file tree

Hide file tree

Showing 12 changed files with 24 additions and 9 deletions.
diff --git a/phylogenetic/config/auspice_config_all.json b/phylogenetic/config/auspice_config_all.json
@@ -3,7 +3,7 @@
   "maintainers": [
     {"name": "Trevor Bedford", "url": "http://bedford.io/team/trevor-bedford/"}
   ],
-  "build_url": "https://github.com/nextstrain/dengue",  
+  "build_url": "https://github.com/nextstrain/dengue",
   "colorings": [
     {
       "key": "gt",
@@ -44,5 +44,8 @@
     "country",
     "region",
     "author"
+  ],
+  "metadata_columns": [
+    "genbank_accession"
   ]
 }
diff --git a/phylogenetic/config/auspice_config_denv1.json b/phylogenetic/config/auspice_config_denv1.json
@@ -43,5 +43,8 @@
     "country",
     "region",
     "author"
+  ],
+  "metadata_columns": [
+    "genbank_accession"
   ]
 }
diff --git a/phylogenetic/config/auspice_config_denv2.json b/phylogenetic/config/auspice_config_denv2.json
@@ -43,5 +43,8 @@
     "country",
     "region",
     "author"
+  ],
+  "metadata_columns": [
+    "genbank_accession"
   ]
 }
diff --git a/phylogenetic/config/auspice_config_denv3.json b/phylogenetic/config/auspice_config_denv3.json
@@ -43,5 +43,8 @@
     "country",
     "region",
     "author"
+  ],
+  "metadata_columns": [
+    "genbank_accession"
   ]
 }
diff --git a/phylogenetic/config/auspice_config_denv4.json b/phylogenetic/config/auspice_config_denv4.json
@@ -43,5 +43,8 @@
     "country",
     "region",
     "author"
+  ],
+  "metadata_columns": [
+    "genbank_accession"
   ]
 }
diff --git a/phylogenetic/config/config_dengue.yaml b/phylogenetic/config/config_dengue.yaml
@@ -1,4 +1,4 @@
-strain_id_field: "accession"
+strain_id_field: "genbank_accession"
 display_strain_field: "strain"
 
 filter:

diff --git a/phylogenetic/example_data/metadata_all.tsv b/phylogenetic/example_data/metadata_all.tsv
@@ -1,4 +1,4 @@
-strain	virus	accession	date	region	country	division	city	db	segment	authors	url	title	journal	paper_url
+strain	virus	genbank_accession	date	region	country	division	city	db	segment	authors	url	title	journal	paper_url
 DENV1/VIETNAM/BIDV1862/2008	dengue	FJ410220	2008-XX-XX	Southeast Asia	Vietnam	Vietnam	Vietnam	?	genome	Alvarado et al.	?	?	?	?
 DENV3/WALLIS_AND_FUTUNA/1404838/1989	dengue	JQ920487	1989-08-14	?	Wallis And Futuna	French Overseas Collectivity	Wallis And Futuna	?	genome	Aubry et al.	?	?	?	?
 DENV/BRAZIL/SJRP778/2013	dengue	KT438585	2013-02-04	South America	Brazil	Brazil	Brazil	?	genome	Araujo et al.	?	?	?	?

diff --git a/phylogenetic/example_data/metadata_denv1.tsv b/phylogenetic/example_data/metadata_denv1.tsv
@@ -1,4 +1,4 @@
-strain	virus	accession	date	region	country	division	city	db	segment	authors	url	title	journal	paper_url
+strain	virus	genbank_accession	date	region	country	division	city	db	segment	authors	url	title	journal	paper_url
 DENV1/INDIA/237/1962	dengue	JF297572	1962-XX-XX	South Asia	India	India	India	?	genome	Cecilia et al.	?	?	?	?
 DENV1/CHINA/RL6/2013	dengue	KJ470727	2013-09-XX	China	China	China	China	?	genome	Guo et al.	?	?	?	?
 DENV1/THAILAND/AILANDKPPKDV08491NC04528V0L/2007	dengue	JQ993108	2007-XX-XX	Southeast Asia	Thailand	Thailand	Thailand	?	genome	Aldstadt et al.	?	?	?	?

diff --git a/phylogenetic/example_data/metadata_denv2.tsv b/phylogenetic/example_data/metadata_denv2.tsv
@@ -1,4 +1,4 @@
-strain	virus	accession	date	region	country	division	city	db	segment	authors	url	title	journal	paper_url
+strain	virus	genbank_accession	date	region	country	division	city	db	segment	authors	url	title	journal	paper_url
 DENV2/INDONESIA/MKS0084/2007	dengue	KC762660	2007-11-21	Southeast Asia	Indonesia	Makassar	Makassar	?	genome	Bifani et al.	?	?	?	?
 DENV2/TAIWAN/71/2002	dengue	EF016253	2002-XX-XX	China	Taiwan	Taiwan	Taiwan	?	genome	Ke et al.	?	?	?	?
 DENV2/PAKISTAN/8D/2011	dengue	JX042512	2011-XX-XX	West Asia	Pakistan	Pakistan	Pakistan	?	genome	Alvi et al.	?	?	?	?

diff --git a/phylogenetic/example_data/metadata_denv3.tsv b/phylogenetic/example_data/metadata_denv3.tsv
@@ -1,4 +1,4 @@
-strain	virus	accession	date	region	country	division	city	db	segment	authors	url	title	journal	paper_url
+strain	virus	genbank_accession	date	region	country	division	city	db	segment	authors	url	title	journal	paper_url
 DENV3/PUERTO_RICO/BIDV1615/2004	dengue	FJ373302	2004-XX-XX	North America	Puerto Rico	Puerto Rico	Puerto Rico	?	genome	Alvarado et al.	?	?	?	?
 DENV3/THAILAND/BIDV2323/2001	dengue	FJ744734	2001-XX-XX	Southeast Asia	Thailand	Kamphaeng Phet	Kamphaeng Phet	?	genome	Alvarado et al.	?	?	?	?
 DENV3/BANGLADESH/JACOB/NA	dengue	AY656673	?	South Asia	Bangladesh	Bangladesh	Bangladesh	?	genome	Aaskov et al.	?	?	?	?

diff --git a/phylogenetic/example_data/metadata_denv4.tsv b/phylogenetic/example_data/metadata_denv4.tsv
@@ -1,4 +1,4 @@
-strain	virus	accession	date	region	country	division	city	db	segment	authors	url	title	journal	paper_url
+strain	virus	genbank_accession	date	region	country	division	city	db	segment	authors	url	title	journal	paper_url
 DENV4/ECUADOR/FSE2098/2006	dengue	GQ139572	2006-XX-XX	South America	Ecuador	Ecuador	Ecuador	?	genome	Alava et al.	?	?	?	?
 DENV4/BRAZIL/SJRP580/2012	dengue	KP188562	2012-12-10	South America	Brazil	Brazil	Brazil	?	genome	Araujo et al.	?	?	?	?
 DENV4/BRAZIL/GUSP0670/2013	dengue	KP704043	2013-03-17	South America	Brazil	Brazil	Brazil	?	genome	Balarini et al.	?	?	?	?

diff --git a/phylogenetic/rules/prepare_sequences.smk b/phylogenetic/rules/prepare_sequences.smk
@@ -21,8 +21,8 @@ rule download:
         metadata = "data/metadata_{serotype}.tsv.zst"
 
     params:
-        sequences_url = "https://data.nextstrain.org/files/dengue/sequences_{serotype}.fasta.zst",
-        metadata_url = "https://data.nextstrain.org/files/dengue/metadata_{serotype}.tsv.zst"
+        sequences_url = "https://data.nextstrain.org/files/workflows/dengue/sequences_{serotype}.fasta.zst",
+        metadata_url = "https://data.nextstrain.org/files/workflows/dengue/metadata_{serotype}.tsv.zst"
     shell:
         """
         curl -fsSL --compressed {params.sequences_url:q} --output {output.sequences}