From 75a51e6ed7177ad3a7b9dc344c9d6d085b07773a Mon Sep 17 00:00:00 2001
From: Jennifer Chang <jennifer.chang.bioinform@gmail.com>
Date: Thu, 17 Oct 2024 00:45:11 -0700
Subject: [PATCH] Allow more lenient refine rules for the global refine
 parameters

Separate washington-state specific refine parameters from a more permissive
global refine parameters.
---
 phylogenetic/build-configs/washington-state/config.yaml | 3 +++
 phylogenetic/defaults/config.yaml                       | 5 ++++-
 phylogenetic/rules/construct_phylogeny.smk              | 9 ++-------
 3 files changed, 9 insertions(+), 8 deletions(-)

diff --git a/phylogenetic/build-configs/washington-state/config.yaml b/phylogenetic/build-configs/washington-state/config.yaml
index 0b91a50..6268f41 100644
--- a/phylogenetic/build-configs/washington-state/config.yaml
+++ b/phylogenetic/build-configs/washington-state/config.yaml
@@ -19,6 +19,9 @@ subsampling:
   country: --query "country == 'USA' and state not in ['WA', 'CA', 'ID', 'OR', 'NV','AZ','NM', 'CO', 'UT', 'WY', 'MT'] and accession != 'NC_009942'" --group-by state year --subsample-max-sequences 300 --min-length '9800'
   force_include: --exclude-all --include ../nextclade/defaults/include.txt
 
+refine:
+  treetime_params: --coalescent opt --clock-filter-iqd 4 --date-inference marginal --date-confidence
+
 traits:
   metadata_columns: [
     'country',
diff --git a/phylogenetic/defaults/config.yaml b/phylogenetic/defaults/config.yaml
index d723efb..a86b515 100644
--- a/phylogenetic/defaults/config.yaml
+++ b/phylogenetic/defaults/config.yaml
@@ -2,7 +2,7 @@ strain_id_field: "accession"
 # Use 'Egypt 1951' as the reference and root, following Mencattelli et al, 2023
 # https://www.nature.com/articles/s41467-023-42185-7
 reference: "defaults/reference_global.gb"
-root: "AF260968_REF"
+root: "AF260968"
 
 # Sequences must be FASTA and metadata must be TSV
 # Both files must be zstd compressed
@@ -68,6 +68,9 @@ subsampling:
   region: --query "is_lab_host != 'true'" --query-columns is_lab_host:str --min-length '9800' --group-by region year --subsample-max-sequences 3000 --exclude defaults/exclude.txt
   force_include: --exclude-all --include defaults/include.txt
 
+refine:
+  treetime_params: --coalescent opt --date-inference marginal --date-confidence
+
 traits:
   metadata_columns: [
     'region',
diff --git a/phylogenetic/rules/construct_phylogeny.smk b/phylogenetic/rules/construct_phylogeny.smk
index 59fd7be..1c18cae 100644
--- a/phylogenetic/rules/construct_phylogeny.smk
+++ b/phylogenetic/rules/construct_phylogeny.smk
@@ -55,9 +55,7 @@ rule refine:
     params:
         metadata_id_columns = config["strain_id_field"],
         root = config["root"],
-        date_inference = "marginal",
-        coalescent = "opt",
-        clock_filter_iqd = 4,
+        treetime_params = config["refine"]["treetime_params"],
     shell:
         """
         augur refine \
@@ -69,9 +67,6 @@ rule refine:
             --output-node-data {output.node_data} \
             --root {params.root} \
             --timetree \
-            --coalescent {params.coalescent} \
-            --date-confidence \
-            --date-inference {params.date_inference} \
-            --clock-filter-iqd {params.clock_filter_iqd} \
+            {params.treetime_params} \
             2>&1 | tee {log}
         """