Merge branch 'dev' into 'main' - Metaphor v1.7.11

metaphor/__init__.py

@@ -2,7 +2,7 @@
 __email__ = "17276653+vinisalazar@users.noreply.github.com"
 __license__ = "MIT"
 __copyright__ = "The University of Melbourne 2023"
-__version__ = "1.7.10"
+__version__ = "1.7.11"
 
 __doc__ = """
 Metaphor top-level module. This is the entrypoint for the `cli`, `config`, and `workflow` packages.
@@ -13,7 +13,7 @@
 
 # Snakemake wrapper version
 # This should match the latest released tag on: https://github.com/snakemake/snakemake-wrappers
-wrapper_version = "v3.3.3"
+wrapper_version = "v3.4.1"
 wrapper_prefix = "https://github.com/snakemake/snakemake-wrappers/raw/"
 
 github_url = "https://github.com/vinisalazar/metaphor"

metaphor/workflow/envs/python-utils.yaml

@@ -4,7 +4,7 @@ channels:
   - bioconda
 dependencies:
   - biopython
-  - pandas>=2
+  - pandas>=2.1
   - python>=3.9
   - seaborn
   - seqstats  # This has to be here because assembly_report.py runs it from Python

metaphor/workflow/rules/annotation.smk

@@ -78,8 +78,7 @@ rule prokka:
         outfile="output/annotation/prokka/{binning_group}/{bin}/{bin}.fna",
     params:
         outdir=lambda w, output: str(Path(output.outfile).parent),
-        kingdom_cmd=lambda w, input: f"grep '{w.bin}' {input.bin_evals}",
-        args=config["prokka"]["args"],
+        args=config["prokka"]["args"] + " --centre X --compliant",
     wildcard_constraints:
         binning_group="|".join(binning_group_names),
     threads: get_threads_per_task_size("small")
@@ -94,7 +93,9 @@ rule prokka:
     shell:
         """
         # Get kingdom from bin eval file
-        kingdom=$({params.kingdom_cmd} | cut -f 5)
+        bin_clean=$(echo {wildcards.bin} | sed 's/_sub$//g')  # Remove '_sub' from corrected bins
+        kingdom=$(grep $bin_clean {input.bin_evals} | cut -f 5)
+        kingdom=$(echo $kingdom | cut -f 1 -d ' ')
         kingdom=$(echo $kingdom | head -c 1 | tr '[a-z]' '[A-Z]'; echo $kingdom | tail -c +2)
 
         prokka --outdir {params.outdir}     \

metaphor/workflow/rules/common.smk

@@ -496,7 +496,7 @@ def get_prokka_output():
     bins_dict = {}
     for group in binning_group_names:
         bins_dict[group] = glob(
-            f"output/binning/DAS_tool/{group}/DAS_tool_DASTool_bins/*"
+            f"output/binning/DAS_tool/{group}/{group}_DASTool_bins/*"
         )
 
     for group, list_of_bins in bins_dict.items():

metaphor/workflow/rules/qc.smk

@@ -108,6 +108,8 @@ rule fastqc_raw:  # qc on raw, unmerged reads
         html="output/qc/fastqc/{sample}-{unit}-{read}-raw.html",
     params:
         "--quiet",
+    resources:
+        mem_mb=10000 if get_max_mb(0.5) > 10000 else get_max_mb(0.5),
     wildcard_constraints:
         sample="|".join(sample_IDs),
         unit="|".join(unit_names),
@@ -127,6 +129,8 @@ rule fastqc_trimmed:  # qc on trimmed reads
     output:
         zip="output/qc/fastqc/{sample}-{unit}-{read}-trimmed_fastqc.zip",
         html="output/qc/fastqc/{sample}-{unit}-{read}-trimmed.html",
+    resources:
+        mem_mb=10000 if get_max_mb(0.5) > 10000 else get_max_mb(0.5),
     params:
         "--quiet",
     wildcard_constraints:
@@ -151,6 +155,8 @@ rule fastqc_merged:  # qc on trimmed, merged reads
     output:
         zip="output/qc/fastqc/{sample}-{read}-merged_fastqc.zip",
         html="output/qc/fastqc/{sample}-{read}-merged.html",
+    resources:
+        mem_mb=10000 if get_max_mb(0.5) > 10000 else get_max_mb(0.5),
     params:
         "--quiet",
     wildcard_constraints: