fixing longqc and nanoplot, making longqc optional, adding a parameter to...

fixing longqc and nanoplot, making longqc optional, adding a parameter to distinguish between pacbio and nanopore

SUBMIT_CONFIG/slurm/cluster.yaml

@@ -340,7 +340,7 @@ nanoplot_hifi:
 
 longQC_hifi:
     jobname:    GEP.{rule}.{wildcards.sample}.{wildcards.readCounter}.{wildcards.smrtornot}
-    mem_mb:     12000
+    mem_mb:     96000
     time:       "04:00:05"
 
 multiqc_hifi:

Snakefile

@@ -139,7 +139,7 @@ elif set(['sample', 'hifi_reads', 'meryl_kmer_size','trimSMRTbell', 'fastQC']).i
 
 	samples=samples.set_index(['sample','readCounter'])
 	longqc_dir = "tools/LongQC"
-
+	sequencing_type = "pb-sequel" if config["use_pacbio"] else "ont-rapid"
 	
 	ruleAllQCFiles=[]
 	### TODO change this column in a generic QC analysis. (both nanoplot and and longqc running)

configuration/config.yaml

 ### PATH TO WHERE YOU WANT YOUR RESULTS FOR THIS RUN STORED - WILL BE AUTOMATICALLY CREATED IF IT DOESN'T EXIST  ###
-Results: "/home/valentina/Desktop/test_gep/results_test_new"
+Results: "/home/valentina/Desktop/test_gep/results_test_new_11"
 
 ### FULL PATH TO YOUR SAMPLESHEET    ###
-samplesTSV: "/home/valentina/Desktop/test_gep/build_hifi_example.tsv"
+samplesTSV: "/home/valentina/Desktop/test_gep/build_nano_example.tsv"
 
 ### PATH TO DEFINE RESOURCES FILE - DO NOT CHANGE UNLESS YOU WISH TO USE IN A DIFFERENT LOCATION  ####
 resources: "configuration/define_resources.yaml"
 
-### With EAR=True the ERGA Assembly Report is automatically generated from the analysis
-EAR: False
 smudgeplot: False
 kmer_plot_flat: False
+use_pacbio: True
+run_longqc: True
+
+### With EAR=True the ERGA Assembly Report is automatically generated from the analysis
+EAR: False

configuration/define_resources.yaml

@@ -186,3 +186,8 @@ merylUnion_hifi:
     mem_mb:     96000
     time:       "12:00:00"
     threads:    12
+
+install_longqc:
+    mem_mb:     4000
+    time:       "01:00:00"
+    threads:    1

rules/build_hifi.smk

@@ -69,7 +69,7 @@ rule trimSMRTBellAdapters_hifi:
 	log:
 		os.path.join(config['Results'], "0_buildDatabases/{sample}/hifiReads/logs/trimSMRTbell.{readCounter}.log")
 	priority:
-		15
+		2
 	conda:
 		os.path.join(workflow.basedir, "envs/UNZIP_and_QC.yaml")
 	shell:
@@ -110,11 +110,18 @@ rule fastQC_hifi:
 		(fastqc {input} -o {params.folder2out} -t {threads}) &> {log}
 		"""
 
+def longqc_input(wildcards):
+	# if longqc is disactivated by the user, this input function return the output of nanoplot rather than the longQC one
+	if config["run_longqc"]:
+		return os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/QC/longQC/web_summary.html")
+	else:
+		return os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/QC/nanoplot/NanoPlot-report.html")
+
 rule multiQC_hifi:
 	input:
 		lambda wildcards: expand(os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/QC/fastqc/{readCounter}.{smrtornot}_fastqc.html"), sample=wildcards.sample, readCounter=dictReadCounter[wildcards.sample], smrtornot=dictSamples[wildcards.sample][1]),
 		os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/QC/nanoplot/NanoPlot-report.html"),
-		os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/QC/longQC/web_summary.html")
+		longqc_input
 	params:
 		folder2qc=os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/QC/fastqc/"),
 		folder2qc2=os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/QC/nanoplot/"),
@@ -145,7 +152,7 @@ rule merylCount_hifi:
 	log:
 		os.path.join(config['Results'], "0_buildDatabases/{sample}/hifiReads/logs/meryl_hifi_count.{readCounter}.{kmer}.{smrtornot}.log")
 	priority:
-		10
+		1
 	conda:
 		os.path.join(workflow.basedir, "envs/MER_STATS.yaml")
 	shell:
@@ -155,7 +162,9 @@ rule merylCount_hifi:
 
 rule merylUnion_hifi:
 	input:
-		lambda wildcards: expand(os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/merylDb/{readCounter}_hifi_dB.{smrtornot}.{kmer}.meryl/"), sample=wildcards.sample, readCounter=dictReadCounter[wildcards.sample], kmer=dictSamples[wildcards.sample][0], smrtornot=dictSamples[wildcards.sample][1])
+		db = lambda wildcards: expand(os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/merylDb/{readCounter}_hifi_dB.{smrtornot}.{kmer}.meryl/"), sample=wildcards.sample, readCounter=dictReadCounter[wildcards.sample], kmer=dictSamples[wildcards.sample][0], smrtornot=dictSamples[wildcards.sample][1]),
+		output_nanoplot = os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/QC/nanoplot/NanoPlot-report.html"),
+		output_longqc = longqc_input
 	params:
 		kmer = "{kmer}",
 		removeReadDIR_trimmed=os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/temp_trimReads/"),
@@ -170,12 +179,12 @@ rule merylUnion_hifi:
 	log:
 		os.path.join(config['Results'], "0_buildDatabases/{sample}/hifiReads/logs/meryl_hifi_combine.{sample}.{kmer}.log")
 	priority:
-		10
+		1
 	conda:
 		os.path.join(workflow.basedir, "envs/MER_STATS.yaml")
 	shell:
 		"""
-		(meryl union-sum {input} output {output}) &> {log}
+		(meryl union-sum {input.db} output {output}) &> {log}
 		rm -r {params.removeReadDIR_trimmed}
 		rm -r {params.removeReadDIR_unzipped}
 		"""
@@ -194,6 +203,8 @@ rule NanoPlot:
 		time=resource['nanoplot_hifi']['time'],
 	log:
 		os.path.join(config['Results'], "0_buildDatabases/{sample}/hifiReads/logs/nanoplot.log")
+	priority:
+		2
 	conda:
 		os.path.join(workflow.basedir, "envs/UNZIP_and_QC.yaml")
 	shell:
@@ -206,7 +217,8 @@ rule longQC:
 		fastq = lambda wildcards: expand(os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/temp_trimReads/{readCounter}.{smrtornot}.fastq"), sample=wildcards.sample, readCounter=dictReadCounter[wildcards.sample], smrtornot=dictSamples[wildcards.sample][1]),
 		path_lonqqc = os.path.join(longqc_dir,"longQC.py")
 	params:
-		folder2out=os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/QC/longQC/")
+		folder2out=os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/QC/longQC/"),
+		sequencing_type = sequencing_type
 	output:
 		os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/QC/longQC/web_summary.html")
 	threads:
@@ -216,13 +228,14 @@ rule longQC:
 		time=resource['longQC_hifi']['time'],
 	log:
 		os.path.join(config['Results'], "0_buildDatabases/{sample}/hifiReads/logs/longQC.log")
+	priority:
+		2
 	conda:
 		os.path.join(workflow.basedir, "envs/LONGQC.yaml")
 	shell:
 		"""
-		## TODO making either pacbio or nanopore for -x
-		rmdir {params.folder2out}
-		(python tools/LongQC/longQC.py sampleqc -x pb-sequel -o {params.folder2out} {input.fastq}) &> {log}
+		rm -r {params.folder2out}
+		(python tools/LongQC/longQC.py sampleqc -x {params.sequencing_type} -o {params.folder2out} {input.fastq}) &> {log}
 		"""
 
 rule install_longqc: