Include frequencies for N450 build

This commit produces tip-frequencies JSON files for both N450 and genome analyses, but only displays frequencies panel for N450.

Start date of 2000 and end date of –6 months were chosen based on data availability. I increased bandwidth by about 2.5x to account for data quantity relative to temporal spread (compared to the 2y flu-like defaults).

phylogenetic/Snakefile

@@ -1,10 +1,11 @@
 genes = ['N450', 'genome']
 
-configfile: "defaults/config.yaml" 
+configfile: "defaults/config.yaml"
 
 rule all:
     input:
-        auspice_json = expand("auspice/measles_{gene}.json", gene=genes)
+        auspice_json = expand("auspice/measles_{gene}.json", gene=genes),
+        tip_frequencies_json = expand("auspice/measles_{gene}_tip-frequencies.json", gene=genes)
 
 include: "rules/prepare_sequences.smk"
 include: "rules/prepare_sequences_N450.smk"

phylogenetic/defaults/auspice_config_N450.json

@@ -62,6 +62,12 @@
     "country",
     "author"
   ],
+  "panels": [
+    "tree",
+    "map",
+    "entropy",
+    "frequencies"
+  ],  
   "metadata_columns": [
     "author"
   ]

phylogenetic/defaults/config.yaml

@@ -25,5 +25,10 @@ refine:
     clock_filter_iqd: 4
 ancestral:
     inference: "joint"
+tip_frequencies:
+    min_date: "2000-01-01"
+    max_date: "6M"
+    narrow_bandwidth: 0.2
+    wide_bandwidth: 0.6
 export:
     metadata_columns: "strain division location"

phylogenetic/rules/annotate_phylogeny.smk

@@ -39,3 +39,32 @@ rule translate:
             --reference-sequence {input.reference} \
             --output {output.node_data} \
         """
+
+rule tip_frequencies:
+    """
+    Estimating KDE frequencies for tips
+    """
+    input:
+        tree = "results/{gene}/tree.nwk",
+        metadata = "data/metadata.tsv"
+    params:
+        strain_id = config["strain_id_field"],
+        min_date = config["tip_frequencies"]["min_date"],
+        max_date = config["tip_frequencies"]["max_date"],
+        narrow_bandwidth = config["tip_frequencies"]["narrow_bandwidth"],
+        wide_bandwidth = config["tip_frequencies"]["wide_bandwidth"]
+    output:
+        tip_freq = "results/{gene}/tip-frequencies.json"
+    shell:
+        """
+        augur frequencies \
+            --method kde \
+            --tree {input.tree} \
+            --metadata {input.metadata} \
+            --metadata-id-columns {params.strain_id} \
+            --min-date {params.min_date} \
+            --max-date {params.max_date} \
+            --narrow-bandwidth {params.narrow_bandwidth} \
+            --wide-bandwidth {params.wide_bandwidth} \
+            --output {output.tip_freq}
+        """

phylogenetic/rules/export.smk

@@ -33,3 +33,13 @@ rule export:
             --include-root-sequence-inline \
             --output {output.auspice_json}
         """
+
+rule final_tip_frequencies:
+    input:
+        tip_freq = "results/{gene}/tip-frequencies.json",
+    output:
+        tip_freq = "auspice/measles_{gene}_tip-frequencies.json"
+    shell:
+        """
+        cp -f {input.tip_freq} {output.tip_freq}
+        """