nextclade: enable clade proposals

nextstrain · Oct 27, 2024 · b09a501 · b09a501
1 parent c5ae153
commit b09a501
Show file tree

Hide file tree

Showing 5 changed files with 64 additions and 10 deletions.
diff --git a/nextclade/config/auspice_config.json b/nextclade/config/auspice_config.json
@@ -19,6 +19,16 @@
       "title": "Country",
       "type": "categorical"
     },
+    {
+      "key": "subclade",
+      "title": "Subclade",
+      "type": "categorical"
+    },
+    {
+      "key": "proposed_clade",
+      "title": "Subclade proposals",
+      "type": "categorical"
+    },
     {
       "key": "region",
       "title": "Region",
@@ -38,7 +48,8 @@
   "filters": [
     "region",
     "country",
-    "clade_membership"
+    "clade_membership",
+    "subclade"
   ],
   "display_defaults": {
     "color_by": "clade_membership",

diff --git a/nextclade/config/config_dict.yaml b/nextclade/config/config_dict.yaml
@@ -14,6 +14,9 @@ builds:
         subclade:
           url: "seasonal_A-H1N1pdm_HA/main/.auto-generated/subclades.tsv"
           key: "subclade"
+        proposed_clade:
+          url: "seasonal_A-H1N1pdm_HA/main/.auto-generated/subclade_proposals.tsv"
+          key: "proposed_clade"
       refs:
         CY121680: # exclude South Korean genomes because of sequencing artifacts close to the start of HA
           filter: "--min-date 2009 --probabilistic-sampling  --exclude-where country='south_korea' qc.overallStatus='bad' --group-by year --min-length 1500  --subsample-max-sequences 1500"
@@ -24,19 +27,22 @@ builds:
         MW626062:
           filter: "--min-date 2019 --probabilistic-sampling  --exclude-where country='south_korea' qc.overallStatus='bad' --group-by year --min-length 1500 --subsample-max-sequences 1500"
           clade_offset: 0
-          hardmin_date: 2015
+          hardmin_date: 2009
           reference_EPI_ISL: EPI1812046
           reference_strain: A/Wisconsin/588/2019
     na:
       changelog: "seasonal_A-H1N1pdm_NA/main/CHANGELOG.md"
       clade_systems:
         clade:
           url: "seasonal_A-H1N1pdm_NA/main/.auto-generated/subclades.tsv"
+        proposed_clade:
+          url: "seasonal_A-H1N1pdm_NA/main/.auto-generated/subclade_proposals.tsv"
+          key: "proposed_clade"
       refs:
         MW626056:
           filter: "--min-date 2019 --probabilistic-sampling --group-by year region --min-length 1400 --subsample-max-sequences 2000"
           clade_offset: 0
-          hardmin_date: 2015
+          hardmin_date: 2009
           reference_EPI_ISL: EPI1812046
           reference_strain: A/Wisconsin/588/2019
   h3n2:
@@ -51,14 +57,14 @@ builds:
         short-clade:
           url: "seasonal_A-H3N2_HA/main/.auto-generated/clades.tsv"
           key: "short-clade"
-        emerging_subclade:
-          url: "seasonal_A-H3N2_HA/emerging/.auto-generated/subclades.tsv"
-          key: "emerging_subclade"
+        proposed_clade:
+          url: "seasonal_A-H3N2_HA/main/.auto-generated/subclade_proposals.tsv"
+          key: "proposed_clade"
       refs:
         EPI1857216:
           filter: "--min-date 2019 --probabilistic-sampling --group-by year region --min-length 1500 --subsample-max-sequences 2000"
           clade_offset: -17
-          hardmin_date: 2015
+          hardmin_date: 2004
           reference_EPI_ISL: EPI1857216
           reference_strain: A/Darwin/6/2021
         CY163680:
@@ -72,11 +78,14 @@ builds:
       clade_systems:
         clade:
           url: "seasonal_A-H3N2_NA/main/.auto-generated/subclades.tsv"
+        proposed_clade:
+          url: "seasonal_A-H3N2_NA/main/.auto-generated/subclade_proposals.tsv"
+          key: "proposed_clade"
       refs:
         EPI1857215:
           filter: "--min-date 2019 --probabilistic-sampling --group-by year region --min-length 1400 --subsample-max-sequences 1500"
           clade_offset: 4
-          hardmin_date: 2015
+          hardmin_date: 2004
           reference_EPI_ISL: EPI1857215
           reference_strain: A/Darwin/6/2021
   vic:
@@ -88,6 +97,9 @@ builds:
         subclade:
           url: "seasonal_B-Vic_HA/main/.auto-generated/subclades.tsv"
           key: "subclade"
+        proposed_clade:
+          url: "seasonal_B-Vic_HA/main/.auto-generated/subclade_proposals.tsv"
+          key: "proposed_clade"
       refs:
         KX058884:
           filter: "--min-date 2014 --probabilistic-sampling --group-by year --min-length 1500 --subsample-max-sequences 2000"
@@ -100,6 +112,9 @@ builds:
       clade_systems:
         clade:
           url: "seasonal_B-Vic_NA/main/.auto-generated/subclades.tsv"
+        proposed_clade:
+          url: "seasonal_B-Vic_NA/main/.auto-generated/subclade_proposals.tsv"
+          key: "proposed_clade"
       refs:
         CY073894:
           filter: "--min-date 2014 --probabilistic-sampling --group-by year region --min-length 1400 --subsample-max-sequences 2000"

diff --git a/nextclade/dataset_config/h1n1pdm/includes.txt b/nextclade/dataset_config/h1n1pdm/includes.txt
@@ -1,3 +1,4 @@
+A/Bangladesh/3002/2015
 A/Lao/1632/2023
 A/NorthCarolina/6/2023
 A/Victoria/114/2023
@@ -52,3 +53,9 @@ A/Ghana/FS-11-206/2011
 A/Ghana/ARI1181/2011
 A/BurkinaFaso/26/2012
 A/Yopougon/GR276/2012
+A/Perth/103/2015
+A/Fiji/3/2016
+A/Helsinki/2430/2012
+A/Gansu-Ganzhou/SWL34/2012
+A/Brisbane/96/2012
+A/Minnesota/23/2014
diff --git a/nextclade/dataset_config/h3n2/includes.txt b/nextclade/dataset_config/h3n2/includes.txt
@@ -15,3 +15,13 @@ A/Macau/9901287/2023
 A/California/70/2023
 A/Norway/9164/2023
 A/Bahrain/1899/2023
+
+A/Texas/50/2012
+A/Hawaii/22/2012
+A/Southauckland/7/2012
+A/Victoria/361/2011
+A/Brisbane/297/2014
+A/Austria/93/2014
+A/Southaustralia/2/2013
+A/SouthAfrica/R06488/2017
+A/Bangladesh/3039/2017
diff --git a/nextclade/scripts/merge_jsons.py b/nextclade/scripts/merge_jsons.py
@@ -1,4 +1,5 @@
-import json, argparse
+import json
+import argparse
 
 def get_clade_configs(name):
     return {
@@ -16,7 +17,14 @@ def get_clade_configs(name):
         "name": "subclade",
         "displayName": "Subclade",
         "description": "Experimental fine-grained subclade annotation."
-    }}.get(name, {'name':name, "displayName":name, "description":""})
+    },
+    "proposed_clade": {
+        "name": "Subclade proposal",
+        "displayName": "Subclade proposal",
+        "description": "Includes proposals of new subclades. These can change anytime.",
+        "hideInWeb": True
+        }
+    }.get(name, {'name':name, "displayName":name, "description":""})
 
 
 if __name__=="__main__":
@@ -56,6 +64,9 @@ def get_clade_configs(name):
         auspice_json['extensions']['nextclade']["clade_node_attrs"] =  [
             get_clade_configs(c) for c in args.clades if c!='default'
         ]
+    if 'subclade' in args.clades:
+        auspice_json['display_defaults']['color_by'] = 'subclade'
+        auspice_json['display_defaults']['branch_label'] = 'subclade'
 
     with open(args.output_pathogen, 'w') as fh:
         json.dump(pathogen_json, fh, indent=2)