From 684664a882381a63a90c8864302cdbb5f3c3587c Mon Sep 17 00:00:00 2001
From: james hadfield <hadfield.james@gmail.com>
Date: Fri, 22 Nov 2024 14:49:50 +1300
Subject: [PATCH] Update max-sequences filtering parameter

It makes more sense to specify this as a filtering parameter. We could
continue using a value which can't be changed according to wildcards
(e.g. `target_sequences_per_tree: 3000`) however by using the
"*/*/*: 3000" syntax we make it clearer that it's possible to make
this specific to certain builds.

The new syntax makes this trivial to implement using a
---
 Snakefile                        |  2 +-
 config/gisaid.yaml               |  4 +++-
 config/h5n1-cattle-outbreak.yaml | 11 +++++------
 3 files changed, 9 insertions(+), 8 deletions(-)

diff --git a/Snakefile b/Snakefile
index 90f467b..ca23abe 100755
--- a/Snakefile
+++ b/Snakefile
@@ -413,7 +413,7 @@ def _filter_params(wildcards, input, output, threads, resources):
     group_by_value = resolve_config_value(['filter', 'group_by'], wildcards)
     cmd += f" --group-by {group_by_value}" if group_by_value else ""
 
-    cmd += f" --subsample-max-sequences {config['target_sequences_per_tree']}"
+    cmd += f" --subsample-max-sequences {resolve_config_value(['filter', 'target_sequences_per_tree'], wildcards)}"
     cmd += f" --min-date {resolve_config_value(['filter', 'min_date'], wildcards)}"
     cmd += f" --include {input.include}"
     cmd += f" --exclude-where {exclude_where}"
diff --git a/config/gisaid.yaml b/config/gisaid.yaml
index 0ab15b3..779d2df 100644
--- a/config/gisaid.yaml
+++ b/config/gisaid.yaml
@@ -46,7 +46,6 @@ subtype_lookup:
   h9n2: ['h9n2']
 
 #### Parameters which control large overarching aspects of the build
-target_sequences_per_tree: 3000
 same_strains_per_segment: false
 
 
@@ -68,6 +67,9 @@ description: config/description_gisaid.md
 # There's one exception: If a config value is constant for any and all builds then you
 # can just use a scalar value (number, string, boolean)
 filter: 
+  target_sequences_per_tree:
+    "*/*/*": 3000
+
   min_length:
     "*/pb2/*": 2100
     "*/pb1/*": 2100
diff --git a/config/h5n1-cattle-outbreak.yaml b/config/h5n1-cattle-outbreak.yaml
index a90474e..9981f21 100644
--- a/config/h5n1-cattle-outbreak.yaml
+++ b/config/h5n1-cattle-outbreak.yaml
@@ -37,12 +37,6 @@ local_ingest: false
 subtype_lookup:
   h5n1-cattle-outbreak: ['h5n1', 'h5n2', 'h5n3', 'h5n4', 'h5n5', 'h5n6', 'h5n7', 'h5n8', 'h5n9']
 
-#### Parameters which control large overarching aspects of the build
-# Set a high target_sequences_per_tree to capture all circulating strains, as they will be pruned down
-# as part of the workflow
-target_sequences_per_tree: 10_000
-
-
 #### Config files ####
 reference: config/h5n1/reference_h5n1_{segment}.gb  # use H5N1 references
 genome_reference: config/{subtype}/h5_cattle_genome_root.gb
@@ -58,6 +52,11 @@ description: config/{subtype}/description_{subtype}.md
 
 #### Rule-specific parameters ####
 filter:
+  # Set a high target_sequences_per_tree to capture all circulating strains, as they will be pruned down
+  # as part of the workflow
+  target_sequences_per_tree:
+    "*/*/*": 3000
+
   min_length:
     "*/pb2/*": 2100 # Note: could use "h5n1-cattle-outbreak/pb2/default: 2100" if desired
     "*/pb1/*": 2100