From 1316da0474dcfae386af2ad0233063bbb1dc6b4c Mon Sep 17 00:00:00 2001
From: farchaab <farid.chaabane@chuv.ch>
Date: Fri, 6 Dec 2024 12:55:57 +0100
Subject: [PATCH] remove tax_collapse

---
 zamp/DBprocess.Snakefile                      |  5 +--
 .../DB_processing/DB_preprocessing.rules      | 33 +++++++++----------
 zamp/utils.py                                 |  7 ----
 3 files changed, 19 insertions(+), 26 deletions(-)

diff --git a/zamp/DBprocess.Snakefile b/zamp/DBprocess.Snakefile
index 530f1dd..45941d0 100644
--- a/zamp/DBprocess.Snakefile
+++ b/zamp/DBprocess.Snakefile
@@ -42,7 +42,6 @@ TAXONOMY = config.args.taxonomy
 CLASSIFIERS = config.args.classifier
 RDP_MEM = config.args.rdp_mem
 PROCESS = config.args.processing
-TAX_COLLAPSE = dict(config.args.tax_collapse)
 
 ## Cutadapt args
 ERRORS = config.args.errors
@@ -72,7 +71,9 @@ if "--use-singularity" in sys.argv:
     workflow.deployment_settings.apptainer_args += (
         f" -B {workflow.basedir}:{workflow.basedir}"
     )
-    workflow.deployment_settings.apptainer_args += f" -B {os.path.abspath(DBPATH)}:{os.path.abspath(DBPATH)}"
+    workflow.deployment_settings.apptainer_args += (
+        f" -B {os.path.abspath(DBPATH)}:{os.path.abspath(DBPATH)}"
+    )
 #### Load a dictionnary of singularity containers that will be called from each rule
 singularity_envs = yaml.safe_load(
     open(os.path.join(workflow.basedir, "envs", "singularity", "sing_envs.yml"), "r")
diff --git a/zamp/rules/DB_processing/DB_preprocessing.rules b/zamp/rules/DB_processing/DB_preprocessing.rules
index 8bb4d11..1ed186a 100644
--- a/zamp/rules/DB_processing/DB_preprocessing.rules
+++ b/zamp/rules/DB_processing/DB_preprocessing.rules
@@ -21,15 +21,15 @@ rule Extract_amplicons_cutadapt:
     threads: 3
     shell:
         """
-        cutadapt \
-        --cores {threads} \
-        --error-rate {params.errors} \
-        --times 1 \
-        -o {output} \
-        -g "{params.adapter}" \
-        --discard-untrimmed \
-        --minimum-length {params.minlen} \
-        --maximum-length {params.maxlen} \
+        cutadapt \\
+        --cores {threads} \\
+        --error-rate {params.errors} \\
+        --times 1 \\
+        -o {output} \\
+        -g "{params.adapter}" \\
+        --discard-untrimmed \\
+        --minimum-length {params.minlen} \\
+        --maximum-length {params.maxlen} \\
         {input} > {log}
         """
 
@@ -49,10 +49,10 @@ rule Dereplicate_amplicons:
     threads: 1
     shell:
         """
-        vsearch \
-        --derep_fulllength {input} \
-        --output {output.fa} \
-        --uc {output.uc} \
+        vsearch \\
+        --derep_fulllength {input} \\
+        --output {output.fa} \\
+        --uc {output.uc} \\
         2> {log}
         """
 
@@ -66,14 +66,13 @@ rule Derep_and_merge_taxonomy:
         tax=os.path.join("{prefix}", "master", "original_tax.txt"),
         uc=os.path.join("{prefix}", "QIIME", "DB_amp.uc"),
     output:
-        formatted_tax=os.path.join("{prefix}", "QIIME", "DB_amp_taxonomy.txt"),
+        collapsed=os.path.join("{prefix}", "QIIME", "DB_amp_taxonomy.txt"),
         all=os.path.join("{prefix}", "QIIME", "DB_amp_all_taxonomy.txt"),
-        problematic=os.path.join("{prefix}", "QIIME", "problematic_taxa.txt"),
+        ambiguous=os.path.join("{prefix}", "QIIME", "ambiguous_taxa.txt"),
     log:
         os.path.join("{prefix}", "logs", "QIIME", "derep_and_merge.log"),
     params:
         db_name=DBNAME,
-        tax_collapse=TAX_COLLAPSE,
     threads: 1
     script:
         os.path.join("scripts", "tax_formatting.py")
@@ -83,7 +82,7 @@ rule Hash_derep_tax:
     input:
         os.path.join("{prefix}", "QIIME", "DB_amp_taxonomy.txt"),
         os.path.join("{prefix}", "QIIME", "DB_amp_all_taxonomy.txt"),
-        os.path.join("{prefix}", "QIIME", "problematic_taxa.txt"),
+        os.path.join("{prefix}", "QIIME", "ambiguous_taxa.txt"),
         os.path.join("{prefix}", "QIIME", "DB_amp.fasta"),
     output:
         os.path.join("{prefix}", "QIIME", "DB_formatted.hash"),
diff --git a/zamp/utils.py b/zamp/utils.py
index 6b236fd..b9ffa69 100644
--- a/zamp/utils.py
+++ b/zamp/utils.py
@@ -224,13 +224,6 @@ def db_options(func):
             help="Extract amplicon regions and merge taxonomy",
             show_default=True,
         ),
-        click.option(
-            "--tax-collapse",
-            help="Dictionary of number of ranks to print limit when collapsing names ",
-            default='{"Species": 5, "Genus": 6}',
-            callback=validate_dict,
-            show_default=True,
-        ),
         click.option(
             "--fw-primer",
             type=str,