Merge pull request #42 from nextstrain/frequencies

Include frequencies for N450 build

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

@@ -26,5 +26,10 @@ 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/export.smk

@@ -35,3 +35,32 @@ rule export:
             --output {output.auspice_json} \
             --description {input.description}
         """
+
+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 = "auspice/measles_{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}
+        """