From 0bd64d3c52d52f538160a9713e50efc2260d5f6d Mon Sep 17 00:00:00 2001
From: Jover Lee <joverlee521@gmail.com>
Date: Wed, 20 Nov 2024 16:21:06 -0800
Subject: [PATCH] phylogenetic: Add config option `ancestral_root_seq`

Allow users to define an optional `--root-sequence` for
`augur ancestral`.

This commit only adds the new config option to the CI config.
It is not used in the default Nextstrain builds because it's not clear
whether it's needed.
---
 phylogenetic/build-configs/ci/config.yaml | 1 +
 phylogenetic/rules/annotate_phylogeny.smk | 8 +++++++-
 2 files changed, 8 insertions(+), 1 deletion(-)

diff --git a/phylogenetic/build-configs/ci/config.yaml b/phylogenetic/build-configs/ci/config.yaml
index f58e32f..1bfdef7 100644
--- a/phylogenetic/build-configs/ci/config.yaml
+++ b/phylogenetic/build-configs/ci/config.yaml
@@ -2,6 +2,7 @@ custom_rules:
   - build-configs/ci/copy_example_data.smk
 
 reference: "defaults/reference.fasta"
+ancestral_root_seq: "defaults/reference.fasta"
 genome_annotation: "defaults/genome_annotation.gff3"
 genbank_reference: "defaults/reference.gb"
 include: "defaults/hmpxv1/include.txt"
diff --git a/phylogenetic/rules/annotate_phylogeny.smk b/phylogenetic/rules/annotate_phylogeny.smk
index 9dea6a3..ccb2414 100644
--- a/phylogenetic/rules/annotate_phylogeny.smk
+++ b/phylogenetic/rules/annotate_phylogeny.smk
@@ -31,13 +31,19 @@ rule ancestral:
         node_data=build_dir + "/{build_name}/nt_muts.json",
     params:
         inference="joint",
+        root_sequence=lambda w: (
+            f"--root-sequence {config['ancestral_root_seq']!r}"
+            if config.get("ancestral_root_seq")
+            else ""
+        ),
     shell:
         """
         augur ancestral \
             --tree {input.tree} \
             --alignment {input.alignment} \
             --output-node-data {output.node_data} \
-            --inference {params.inference}
+            --inference {params.inference} \
+            {params.root_sequence}
         """