From f152ce213912ec631e3af442771427071db5773d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Karel=20B=C5=99inda?= <karel.brinda@gmail.com>
Date: Wed, 6 Mar 2024 12:21:28 +0100
Subject: [PATCH] Make attotree read the list of input files from a file

---
 workflow/rules/asm.smk  | 17 ++++++++++++++++-
 workflow/rules/init.smk | 11 +++++++++++
 workflow/rules/tree.smk |  4 ++--
 3 files changed, 29 insertions(+), 3 deletions(-)

diff --git a/workflow/rules/asm.smk b/workflow/rules/asm.smk
index 8ee354e..ed82906 100644
--- a/workflow/rules/asm.smk
+++ b/workflow/rules/asm.smk
@@ -3,7 +3,22 @@
 ##
 
 
-rule asm_list:
+rule asm_list_unsorted:
+    """
+    Make an unsorted list of input assemblies
+    """
+    output:
+        list=fn_asm_list_unsorted(_batch="{batch}"),
+    input:
+        fa=w_batch_asms,
+    run:
+        dump_file_list(
+            input.fa,
+            output.list,
+        )
+
+
+rule asm_list_sorted:
     """
     Make a list of assemblies as they will appear in the .tar.xz archive
     """
diff --git a/workflow/rules/init.smk b/workflow/rules/init.smk
index c0609be..d149668 100644
--- a/workflow/rules/init.smk
+++ b/workflow/rules/init.smk
@@ -148,6 +148,10 @@ def fn_hist_summary(_batch, _protocol):
 # ASM
 
 
+def fn_asm_list_unsorted(_batch):
+    return f"{dir_intermediate()}/asm/{_batch}.asm.list_unsorted"
+
+
 def fn_asm_seq_dir(_batch):
     return f"{dir_intermediate()}/asm/{_batch}"
 
@@ -235,6 +239,13 @@ def generate_file_list(input_list_fn, output_list_fn, filename_function):
                 g.write(fn + "\n")
 
 
+# dump list of files into a file
+def dump_file_list(input_files, output_list_fn):
+    with open(output_list_fn, "w") as g:
+        for x in input_files:
+            g.write(x + "\n")
+
+
 def load_list(fn):
     try:
         with open(fn) as f:
diff --git a/workflow/rules/tree.smk b/workflow/rules/tree.smk
index 87e9f4e..79ce4c7 100644
--- a/workflow/rules/tree.smk
+++ b/workflow/rules/tree.smk
@@ -71,7 +71,7 @@ if not config["trees_required"]:
         output:
             nw=fn_tree_dirty(_batch="{batch}"),
         input:
-            w_batch_asms,
+            nw=fn_asm_list_unsorted(_batch="{batch}"),
         threads: config["attotree_threads"]
         params:
             k=config["attotree_kmer_length"],
@@ -85,6 +85,6 @@ if not config["trees_required"]:
                 -t {params.t} \\
                 -k {params.k} \\
                 -s {params.s} \\
-                {input} \\
+                -L {input.nw} \\
                 | tee {output.nw}
             """