From 82d5231b5a640104a136abffce353686d4e68b03 Mon Sep 17 00:00:00 2001
From: skchronicles <kuhnsa3@gmail.com>
Date: Mon, 8 Apr 2024 16:10:57 -0400
Subject: [PATCH] Adding option to add branch supports via bootstrapping

---
 mpox-seek               | 21 ++++++++++++++++++++-
 workflow/Snakefile      |  4 ++++
 workflow/rules/tree.smk |  3 ++-
 3 files changed, 26 insertions(+), 2 deletions(-)

diff --git a/mpox-seek b/mpox-seek
index c35691d..c1425c8 100755
--- a/mpox-seek
+++ b/mpox-seek
@@ -399,6 +399,7 @@ def parsed_arguments(name, description):
                 [--conda-env-name CONDA_ENV_NAME] \\
                 [--additional-strains ADDITIONAL_STRAINS] \\
                 [--batch-id BATCH_ID] \\
+                [--bootstrap-trees] \\
                 --input INPUT [INPUT ...] \\
                 --output OUTPUT
 
@@ -444,7 +445,6 @@ def parsed_arguments(name, description):
                                 the option listed below.
                                   Example:
                                     --additional-strains resources/mpox_additional_strains.fa.gz
-
           --batch-id BATCH_ID
                                 Unique identifer to associate with a batch of samples.
                                 This option can be provided to ensure that project-level
@@ -462,6 +462,13 @@ def parsed_arguments(name, description):
                                 Here is a list of valid or acceptable characters: "aA-Zz",
                                 "0-9", "-", "_".
                                   Example: --batch-id "2024-04-01"
+          --bootstrap-trees
+                                Computes branch support by bootstraping data. If this flag
+                                is provided, raxml-ng is run in an all-in-one (ML search + 
+                                bootstrapping) mode via its '--all' option. Branch supports,
+                                calculated by bootstrapping, will be added to the best 
+                                scoring tree. 
+                                  Example: --bootstrap-trees
 
         {3}{4}Orchestration options:{5}
           --mode {{local,slurm}}  
@@ -593,6 +600,7 @@ def parsed_arguments(name, description):
                     --output mpox-seek_output \\
                     --additional-strains resources/mpox_additional_strains.fa.gz \\
                     --batch-id "$(date '+%Y-%m-%d-%H-%M')" \\
+                    --bootstrap-trees \\
                     --mode local \\
                     --use-conda \\
                     --conda-env-name mpox-seek \\
@@ -609,6 +617,7 @@ def parsed_arguments(name, description):
                     --output mpox-seek_output \\
                     --additional-strains resources/mpox_additional_strains.fa.gz \\
                     --batch-id "$(date '+%Y-%m-%d-%H-%M')" \\
+                    --bootstrap-trees \\
                     --use-conda \\
                     --conda-env-name mpox-seek \\
                     --mode local
@@ -666,6 +675,16 @@ def parsed_arguments(name, description):
         help = argparse.SUPPRESS
     )
 
+    # Add branch support/confidence values
+    # to best tree via bootstrapping
+    subparser_run.add_argument(
+        '--bootstrap-trees',
+        action = 'store_true',
+        required = False,
+        default = False,
+        help = argparse.SUPPRESS
+    )
+
     # Optional Arguments
     # Add custom help message
     subparser_run.add_argument(
diff --git a/workflow/Snakefile b/workflow/Snakefile
index 2e61a78..7f2c1e9 100644
--- a/workflow/Snakefile
+++ b/workflow/Snakefile
@@ -37,6 +37,10 @@ decompress_strains_fasta = False
 if strains_fasta.endswith('.gz') or strains_fasta.endswith('.gzip'):
     decompress_strains_fasta = True 
 
+# Maps branch support/confidence values
+# to the best tree via bootstrapping data 
+bootstrap_trees = str_bool(config['options']['bootstrap_trees'])
+
 # Find list of sample which 
 # have mulitple barcodes, this 
 # means they need to be merged  
diff --git a/workflow/rules/tree.smk b/workflow/rules/tree.smk
index 79c39d0..a619dee 100644
--- a/workflow/rules/tree.smk
+++ b/workflow/rules/tree.smk
@@ -18,6 +18,7 @@ rule tree:
     params:
         rname  = 'tree',
         prefix = join(workpath, "project", batch_id, "mpox_phylogeny"),
+        bootrapping_options = lambda _: "--all --bs-metric fbp,tbe" if bootstrap_trees else ""
     conda: depending(conda_yaml_or_named_env, use_conda)
     container: depending(config['images']['mpox-seek'], use_singularity)
     shell: 
@@ -31,5 +32,5 @@ rule tree:
             --model GTR+G \\
             --msa-format FASTA \\
             --prefix {params.prefix} \\
-            --seed 42
+            --seed 42 {params.bootrapping_options}
         """
\ No newline at end of file