some fixes related to scaffold and ccs

sequana_pipelines/lora/config.yaml

@@ -31,7 +31,7 @@ apptainers:
     medaka: https://zenodo.org/record/7846295/files/medaka_1.7.3.img
     miniasm: https://zenodo.org/record/13550867/files/miniasm_0.3.0.img
     mosdepth: https://zenodo.org/record/12805460/files/mosdepth_0.3.8.img
-    multiqc: https://zenodo.org/record/10205070/files/multiqc_1.16.0.img
+    multiqc: https://zenodo.org/record/17100751/files/multiqc_1.27.0-zenodo1.img
     necat: https://zenodo.org/record/13907089/files/necat_0.0.1.img
     pecat: https://zenodo.org/record/13942767/files/pecat_0.0.3.img
     pbindex: https://zenodo.org/record/7817189/files/pbbam_2.3.0.img
@@ -463,7 +463,7 @@ fastp:
     min_length_required: 1000
     threads: 4
     resources:
-        mem: 4G
+        mem: 8G
 
 
 

sequana_pipelines/lora/main.py

@@ -168,6 +168,7 @@ def convert(self, value, param, ctx):
     "--do-circlator",
     "circlator",
     default=False,
+    is_flag=True,
     help="Run circlator after assembler. Useful with Canu, Hifiasm. Flye is suppose to perform circularisation.",
 )
 @click.option("--do-coverage", "do_coverage", is_flag="store_true", help="Run sequana coverage on contigs.")
@@ -176,7 +177,7 @@ def convert(self, value, param, ctx):
     "--busco-lineage",
     "lineage",
     type=BUSCO_OR_DIR,
-    help="Lineage or path to lineage file for BUSCO. Note that we support only version 5 of the BUSCO lineage. If the lineage is not a valid path, we download it locally. This may take time. Full list is on https://github.com/sequana/lora//wiki/busco",
+    help="Lineage or path to lineage file for BUSCO. Note that we support only version 5 of the BUSCO lineage. If the lineage is not a valid path, we download it locally. The full list is available on https://github.com/sequana/lora/wiki/busco . For bacteria, you may simply use 'bacteria'",
 )
 @click.option(
     "--checkm-rank",
@@ -353,7 +354,7 @@ def main(**options):
     cfg.circlator["do"] = True if options.circlator else False
     if options.assembler == "canu" and options.circlator is False:
         logger.warning(
-            "\U0001F449 \u2757 Canu would higly benefit for circlator. However, you did not set it. please conider using --circlator."
+            "\U0001F449 \u2757 Canu would higly benefit for circlator. However, you did not set it. please conider using --do-circlator."
         )
 
     # blast
@@ -407,11 +408,11 @@ def main(**options):
 
     def _convert(x):
         if x.endswith("k"):
-            x = int(x.strip("k")) * 1000
+            x = int(float(x.strip("k")) * 1000)
         elif x.endswith("m"):
-            x = int(x.strip("m")) * 1000000
+            x = int(float(x.strip("m")) * 1000000)
         elif x.endswith("g"):
-            x = int(x.strip("g")) * 1000000000
+            x = int(float(x.strip("g")) * 1000000000)
         else:
             x = int(x.strip())
         return x

sequana_pipelines/lora/rules/ccs.smk

@@ -132,7 +132,16 @@ rule fastp:
         "{sample}/logs/fastp.log",
     shell:
         """
-        fastp --in1 {input} --length_required {params.min_length_required} --out1 {output} --disable_adapter_trimming --disable_quality_filtering --html {output.html} --json {output.json} --thread {threads} 2>&1 > {log}
+        # reduce RAM usage by first streaming from disk instead of fully loading
+        # bam_to_fastq does not compress the data
+
+        if [[ "{input}" == *.gz ]]; then
+            catcmd="zcat"
+        else
+            catcmd="cat"
+        fi
+
+        $catcmd  {input} | fastp -i /dev/stdin --length_required {params.min_length_required} --out1 {output} --disable_adapter_trimming --disable_quality_filtering --html {output.html} --json {output.json} --thread {threads} 2>&1 > {log}
         """
 
 

sequana_pipelines/lora/rules/scaffold.smk

@@ -17,6 +17,6 @@ rule ragtag_scaffold:
         config['apptainers']['ragtag']
     shell:
         """
-        ragtag_scaffold.py {input.ctg} {input.ref} -w -o {wildcards.sample}/ragtag_scaffold 2>{log}
+        ragtag_scaffold.py {input.ref} {input.ctg} -w -o {wildcards.sample}/ragtag_scaffold 2>{log}
         """
 

sequana_pipelines/lora/src/utils.py

@@ -49,6 +49,7 @@ def checkm_version_parser(stdout: str):
     1: {
         "seqtk": seqtk_version_parser,
     },
+    255: {},
 }
 APPTAINER_CORRESPONDANCE = {
     "canu_correction": "canu",