working H7Nx,HxN6 builds

nextstrain · Dec 4, 2024 · 04f527d · 04f527d
1 parent 4525438
commit 04f527d
Show file tree

Hide file tree

Showing 11 changed files with 272 additions and 39 deletions.
diff --git a/Snakefile b/Snakefile
@@ -582,6 +582,7 @@ rule refine:
     params:
         coalescent = config['refine']['coalescent'],
         date_inference = config['refine']['date_inference'],
+        timetree = lambda w: "--timetree" if resolve_config_value(['refine', 'timetree'], w) else "",
         clock_rates = refine_clock_rates,
         clock_filter = refine_clock_filter,
         root = lambda w: f"--root {resolve_config_value(['refine', 'root'], w)}" if resolve_config_value(['refine', 'root'], w) else ''
@@ -593,7 +594,7 @@ rule refine:
             --metadata {input.metadata} \
             --output-tree {output.tree} \
             --output-node-data {output.node_data} \
-            --timetree \
+            {params.timetree} \
             {params.root} \
             --coalescent {params.coalescent} \
             --date-confidence \

diff --git a/config/h7n6.yaml b/config/h7n6.yaml
@@ -5,57 +5,23 @@ inputs:
 
 builds:
   - subtype: h7n6
-    segment:
-      - pb2
-      - pb1
-      - pa
-      - ha
-      - np
-      - na
-      - mp
-      - ns
+    segment: ['pb2', 'pb1', 'pa', 'ha', 'np', 'na', 'mp', 'ns']
     time: all-time
 
 subtype_lookup:
   h7n6: ['h7n6']
 
-
-
-#### Config files ####
-reference: config/{subtype}/reference_{subtype}_{segment}.gb
-auspice_config: config/{subtype}/auspice_config_{subtype}.json
-colors: config/{subtype}/colors_{subtype}.tsv
-lat_longs: config/{subtype}/lat_longs_{subtype}.tsv
-include_strains: config/{subtype}/include_strains_{subtype}_{time}.txt
-dropped_strains: config/{subtype}/dropped_strains_{subtype}.txt
-description: config/description_gisaid.md
-
-
 filter: 
-  target_sequences_per_tree:
-    "*/*/*": 3000
-
-  min_length:
-    "*/pb2/*": 2100
-    "*/pb1/*": 2100
-    "*/pa/*": 2000
-    "*/ha/*": 1600
-    "*/np/*": 1400
-    "*/na/*": 1270
-    "*/mp/*": 900
-    "*/ns/*": 800
-
   min_date:
     "h7n6/*/all-time": 1900
   group_by:
     "h7n6/*/*": subtype
   exclude_where:
     host=laboratoryderived # can't use an empty string...
 
-ancestral:
-  inference: joint
-  root_seq: false
-
+refine:
+  clock_filter_iqd:
+    "h7n6/*/*": ""
 traits:
   columns:
     "h7n6/*/*": region
diff --git a/config/h7nx/auspice_config.json b/config/h7nx/auspice_config.json
@@ -0,0 +1,97 @@
+{
+  "title": "Real-time tracking of influenza A/H7Nx evolution",
+  "maintainers": [
+    {"name": "James Hadfield"},
+    {"name": "Nextstrain"}
+  ],
+  "build_url": "https://github.com/nextstrain/avian-flu",
+  "data_provenance": [
+    {
+      "name": "GISAID"
+    }
+  ],
+  "colorings": [
+    {
+      "key": "gt",
+      "title": "Genotype",
+      "type": "categorical"
+    },
+    {
+      "key": "num_date",
+      "title": "Date",
+      "type": "continuous"
+    },
+    {
+      "key": "date",
+      "title": "Date",
+      "type": "temporal"
+    },
+    {
+      "key": "pathogenicity",
+      "title": "Pathogenicity",
+      "type": "categorical"
+    },
+    {
+      "key": "subtype",
+      "title": "subtype",
+      "type": "categorical"
+    },
+    {
+      "key": "region",
+      "title": "Region",
+      "type": "categorical"
+    },
+    {
+      "key": "country",
+      "title": "Country",
+      "type": "categorical"
+    },
+    {
+      "key": "division",
+      "title": "Admin Division",
+      "type": "categorical"
+    },
+    {
+      "key": "host",
+      "title": "Host",
+      "type": "categorical"
+    },
+    {
+      "key": "furin_cleavage_motif",
+      "title": "Furin Cleavage Motif",
+      "type": "categorical"
+    },
+    {
+      "key": "cleavage_site_sequence",
+      "title": "Cleavage Site Sequence",
+      "type": "categorical"
+    },
+    {
+      "key": "originating_lab",
+      "title": "Originating Lab",
+      "type": "categorical"
+    },
+    {
+      "key": "submitting_lab",
+      "title": "Submitting Lab",
+      "type": "categorical"
+    }
+  ],
+  "geo_resolutions": [
+    "region",
+    "country",
+    "division"
+  ],
+  "display_defaults": {
+    "map_triplicate": true,
+    "color_by": "pathogenicity",
+    "geo_resolution": "region"
+  },
+  "filters": [
+    "host",
+    "country",
+    "division",
+    "originating_lab",
+    "submitting_lab"
+  ]
+}
diff --git a/config/h7nx/dropped_strains.txt b/config/h7nx/dropped_strains.txt
@@ -0,0 +1,2 @@
+A/duck/Chiba/13/200
+
diff --git a/config/h7nx/include_strains.txt b/config/h7nx/include_strains.txt
diff --git a/config/hxn6/auspice_config.json b/config/hxn6/auspice_config.json
@@ -0,0 +1,97 @@
+{
+  "title": "Real-time tracking of influenza A/HxN6 evolution",
+  "maintainers": [
+    {"name": "James Hadfield"},
+    {"name": "Nextstrain"}
+  ],
+  "build_url": "https://github.com/nextstrain/avian-flu",
+  "data_provenance": [
+    {
+      "name": "GISAID"
+    }
+  ],
+  "colorings": [
+    {
+      "key": "gt",
+      "title": "Genotype",
+      "type": "categorical"
+    },
+    {
+      "key": "num_date",
+      "title": "Date",
+      "type": "continuous"
+    },
+    {
+      "key": "date",
+      "title": "Date",
+      "type": "temporal"
+    },
+    {
+      "key": "pathogenicity",
+      "title": "Pathogenicity",
+      "type": "categorical"
+    },
+    {
+      "key": "subtype",
+      "title": "subtype",
+      "type": "categorical"
+    },
+    {
+      "key": "region",
+      "title": "Region",
+      "type": "categorical"
+    },
+    {
+      "key": "country",
+      "title": "Country",
+      "type": "categorical"
+    },
+    {
+      "key": "division",
+      "title": "Admin Division",
+      "type": "categorical"
+    },
+    {
+      "key": "host",
+      "title": "Host",
+      "type": "categorical"
+    },
+    {
+      "key": "furin_cleavage_motif",
+      "title": "Furin Cleavage Motif",
+      "type": "categorical"
+    },
+    {
+      "key": "cleavage_site_sequence",
+      "title": "Cleavage Site Sequence",
+      "type": "categorical"
+    },
+    {
+      "key": "originating_lab",
+      "title": "Originating Lab",
+      "type": "categorical"
+    },
+    {
+      "key": "submitting_lab",
+      "title": "Submitting Lab",
+      "type": "categorical"
+    }
+  ],
+  "geo_resolutions": [
+    "region",
+    "country",
+    "division"
+  ],
+  "display_defaults": {
+    "map_triplicate": true,
+    "color_by": "pathogenicity",
+    "geo_resolution": "region"
+  },
+  "filters": [
+    "host",
+    "country",
+    "division",
+    "originating_lab",
+    "submitting_lab"
+  ]
+}
diff --git a/config/hxn6/dropped_strains.txt b/config/hxn6/dropped_strains.txt
diff --git a/config/hxn6/include_strains.txt b/config/hxn6/include_strains.txt
diff --git a/config/hxn6_h7nx.yaml b/config/hxn6_h7nx.yaml
@@ -0,0 +1,64 @@
+inputs:
+  - name: gisaid
+    metadata: ingest/fauna/results/metadata.tsv
+    sequences: ingest/fauna/results/sequences_{segment}.fasta
+
+builds:
+  - subtype:
+      - h7nx
+      - hxn6
+    segment:
+      # - pb2
+      # - pb1
+      # - pa
+      # - ha
+      # - np
+      # - na
+      # - mp
+      - ns
+    time: all-time
+
+## HACK! FIXME XXX
+auspice_config: config/{subtype}/auspice_config.json
+reference: config/h7n9/reference_h7n9_{segment}.gb
+colors: config/h7n6/colors_h7n6.tsv
+lat_longs: config/h7n9/lat_longs_h7n9.tsv
+include_strains: config/{subtype}/include_strains.txt
+dropped_strains: config/{subtype}/dropped_strains.txt
+clades_file: clade-labeling/{subtype}-clades.tsv # TODO XXX -- should drop this! But I think it's required. BUG.
+description: config/description_gisaid.md
+
+### TODO
+# fix the clock?
+
+subtype_lookup:
+  h7nx: ['h7n1', 'h7n2', 'h7n3', 'h7n4', 'h7n5', 'h7n6', 'h7n7', 'h7n8', 'h7n9', 'h7n10', 'h7n11']
+  hxn6: ['h1n6', 'h2n6', 'h3n6', 'h4n6', 'h5n6', 'h6n6', 'h7n6', 'h8n6', 'h9n6', 'h10n6', 'h11n6', 'h12n6', 'h13n6', 'h14n6', 'h15n6', 'h16n6', 'h17n6', 'h18n6']
+
+filter: 
+  target_sequences_per_tree: 10_000 # FIXME XXX - there's circa 7k sequences so we can build em all
+  min_date:
+    "h7nx/*/all-time": 1900
+    "hxn6/*/all-time": 1900
+  min_length:
+    "hxn6/ns/*": 600 # NZ sequence is short here...
+    "h7nx/ns/*": 600 # NZ sequence is short here...
+  group_by:
+    "h7nx/*/*": subtype
+    "hxn6/*/*": subtype
+  exclude_where:
+    host=laboratoryderived # can't use an empty string...
+
+refine:
+  timetree:
+    "h7nx/*/*": false
+    "hxn6/*/*": false
+  root:
+    "h7nx/*/*": mid_point
+    "hxn6/*/*": mid_point
+  # clock_filter_iqd:
+  #   "h7nx/*/*": ""
+traits:
+  columns:
+    "h7nx/*/*": region
+    "hxn6/*/*": region
diff --git a/config/schema.yaml b/config/schema.yaml
@@ -215,7 +215,11 @@ properties:
       - clock_filter_iqd
       - clock_rates
       - root
+      - timetree
     properties:
+      timetree:
+        title: Infer timetree?
+        <<: *wildcard_param_property
       coalescent:
         title: Coalescent time scale
         description: >

diff --git a/gisaid/config.yaml b/gisaid/config.yaml
@@ -96,6 +96,8 @@ filter:
     host=laboratoryderived host=ferret host=unknown host=other host=host country=? region=? gisaid_clade=3C.2
 
 refine:
+  timetree:
+    "*/*/*": true
   coalescent: const
   date_inference: marginal