Merge pull request #17 from nextstrain/fix-clock-rate

phylo: Add fixed clock rate for default builds

Author: joverlee521

.github/ci.yaml

@@ -25,6 +25,13 @@ on:
         description: 'Specific container image to use for phylogenetic workflow (will override the default of "nextstrain build")'
         required: false
         type: string
+      trial_name:
+        description: |
+          Trial name for deploying builds.
+          If not set, builds will overwrite existing builds at s3://nextstrain-data/hmpv*
+          If set, builds will be deployed to s3://nextstrain-staging/hmpv_trials_<trial_name>_*
+        required: false
+        type: string
 
 jobs:
   check-new-data:
@@ -90,11 +97,19 @@ jobs:
       runtime: docker
       env: |
         NEXTSTRAIN_DOCKER_IMAGE: ${{ inputs.image }}
+        TRIAL_NAME: ${{ inputs.trial_name }}
       run: |
+        declare -a config
+
+        if [[ "$TRIAL_NAME" ]]; then
+          config+=("deploy_url=s3://nextstrain-staging/hmpv_trials_${TRIAL_NAME}_")
+        fi
+
         nextstrain build \
           phylogenetic \
             deploy \
-            --configfile defaults/config.yaml
+            --configfile defaults/config.yaml \
+            --config "${config[@]}"
       # Specifying artifact name to differentiate ingest build outputs from
       # the phylogenetic build outputs
       artifact-name: phylogenetic-build-output

.github/fetch-and-ingest.yaml

@@ -38,13 +38,29 @@ filter:
     all: 'subtypes'
 
 files:
+  exclude: "defaults/exclude.txt"
   color_schemes: "defaults/colors.tsv"
   auspice_config: "defaults/auspice_config.json"
 
 refine:
   coalescent: "opt"
   date_inference: "marginal"
   clock_filter_iqd: 4
+  # Estimated clock rates from successful hmpv/a builds as of 2025-11-05:
+  # https://nextstrain.org/hmpv/a/genome@2025-11-05?l=clock
+  clock_rate:
+    a:
+      genome: 0.000833
+      G: 0.00278
+      F: 0.000763
+    b:
+      genome: 0.000833
+      G: 0.00278
+      F: 0.000763
+    all:
+      genome: 0.000833
+      G: 0.00278
+      F: 0.000763
 
 ancestral:
   inference: "joint"

.github/rebuild.yaml

@@ -0,0 +1,2 @@
+	PP_0049YFJ # This was causing all tree to not be rooted properly 
+	PP_0049ZYG # This was causing all tree to not be rooted properly 

.github/workflows/phylogenetic.yaml

@@ -31,6 +31,22 @@ rule tree:
             --output {output.tree}
         """
 
+def _clock_rate_params(wildcards):
+    """
+    Generate the clock rate parameters for augur refine based on
+    wildcard subtype and build values are in the config file
+    """
+    clock_rate = (config['refine']
+        .get('clock_rate', {})
+        .get(wildcards.subtype, {})
+        .get(wildcards.build, None))
+
+    if clock_rate is not None:
+        return f'--clock-rate {clock_rate!r}'
+    else:
+        return ""
+
+
 rule refine:
     """
     Refining tree
@@ -50,6 +66,7 @@ rule refine:
         coalescent = config["refine"]["coalescent"],
         date_inference = config["refine"]["date_inference"],
         clock_filter_iqd = config["refine"]["clock_filter_iqd"],
+        clock_rate = _clock_rate_params,
         strain_id = config.get("strain_id_field", "strain"),
 
     shell:
@@ -65,5 +82,6 @@ rule refine:
             --coalescent {params.coalescent} \
             --date-confidence \
             --date-inference {params.date_inference} \
+            {params.clock_rate} \
             --clock-filter-iqd {params.clock_filter_iqd} 
-        """
+        """

phylogenetic/defaults/config.yaml

@@ -52,7 +52,8 @@ rule filter:
     """
     input:
         sequences = "data/sequences.fasta",
-        metadata = "data/metadata.tsv"
+        metadata = "data/metadata.tsv",
+        exclude = config["files"]["exclude"],
     output:
         sequences = "results/{subtype}/{build}/filtered.fasta",
         metadata = "results/{subtype}/{build}/metadata.tsv",
@@ -76,7 +77,8 @@ rule filter:
             --output-metadata {output.metadata} \
             --group-by {params.group_by} \
             --subsample-max-sequences {params.subsample_max_sequences} \
-            --min-date {params.min_date}
+            --min-date {params.min_date} \
+            --exclude {input.exclude}
         """
 
 rule align: