From 8bef86e88f80cc2a5af2721aadb6c759b1d0875c Mon Sep 17 00:00:00 2001
From: John SJ Anderson <janders4@fredhutch.org>
Date: Wed, 20 Nov 2024 16:22:58 -0800
Subject: [PATCH] Add frequencies panel [#2]

This is _largely_ copy-pasta-ed from the `measles` repo; the one
significant change is I enabled the frequencies on both the whole
genome and prM-E builds, and I set the `min_date` param to
`2017-01-01`, because virtually all of the post-1927 samples are dated
2017 and later.
---
 phylogenetic/Snakefile                        |  1 +
 .../defaults/auspice_config_genome.json       |  6 ++++
 .../defaults/auspice_config_prM-E.json        |  6 ++++
 phylogenetic/defaults/config.yaml             |  5 ++++
 phylogenetic/rules/export.smk                 | 30 +++++++++++++++++++
 5 files changed, 48 insertions(+)

diff --git a/phylogenetic/Snakefile b/phylogenetic/Snakefile
index bd939ce..4e2fcdc 100644
--- a/phylogenetic/Snakefile
+++ b/phylogenetic/Snakefile
@@ -5,6 +5,7 @@ gene = ["genome", "prM-E"]
 rule all:
     input:
         auspice_json=expand("auspice/yellow-fever-virus_{gene}.json", gene=gene),
+        tip_frequencies_json=expand("auspice/yellow-fever-virus_{gene}_tip-frequencies.json", gene=gene),
 
 
 include: "rules/prepare_sequences.smk"
diff --git a/phylogenetic/defaults/auspice_config_genome.json b/phylogenetic/defaults/auspice_config_genome.json
index 67f88f2..3e7221b 100644
--- a/phylogenetic/defaults/auspice_config_genome.json
+++ b/phylogenetic/defaults/auspice_config_genome.json
@@ -58,6 +58,12 @@
     "author",
     "host"
   ],
+  "panels": [
+    "tree",
+    "map",
+    "entropy",
+    "frequencies"
+  ],
   "metadata_columns": [
     "author"
   ]
diff --git a/phylogenetic/defaults/auspice_config_prM-E.json b/phylogenetic/defaults/auspice_config_prM-E.json
index 06791f3..575e3e7 100644
--- a/phylogenetic/defaults/auspice_config_prM-E.json
+++ b/phylogenetic/defaults/auspice_config_prM-E.json
@@ -58,6 +58,12 @@
     "author",
     "host"
   ],
+  "panels": [
+    "tree",
+    "map",
+    "entropy",
+    "frequencies"
+  ],
   "metadata_columns": [
     "author"
   ]
diff --git a/phylogenetic/defaults/config.yaml b/phylogenetic/defaults/config.yaml
index 8046e41..7082318 100644
--- a/phylogenetic/defaults/config.yaml
+++ b/phylogenetic/defaults/config.yaml
@@ -33,5 +33,10 @@ ancestral:
     inference: "joint"
 traits:
     columns: "clade region"
+tip_frequencies:
+    min_date: "2017-01-01"
+    max_date: "6M"
+    narrow_bandwidth: 0.2
+    wide_bandwidth: 0.6
 export:
     metadata_columns: "clade region"
diff --git a/phylogenetic/rules/export.smk b/phylogenetic/rules/export.smk
index d752260..c2038ad 100644
--- a/phylogenetic/rules/export.smk
+++ b/phylogenetic/rules/export.smk
@@ -39,3 +39,33 @@ rule export:
             --description {input.description:q} \
           2> {log:q}
         """
+
+
+rule tip_frequencies:
+    """
+    Estimating KDE frequencies for tips
+    """
+    input:
+        tree = "results/{gene}/tree.nwk",
+        metadata = "data/metadata.tsv"
+    output:
+        tip_freq = "auspice/yellow-fever-virus_{gene}_tip-frequencies.json"
+    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"]
+    shell:
+        r"""
+        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}
+        """