From 43b4c8be9cfe0b549500a0ab23d2e530b6e56e6d Mon Sep 17 00:00:00 2001
From: valegale <valentina_galeone@outlook.it>
Date: Wed, 12 Feb 2025 02:57:28 +0100
Subject: [PATCH] adding longQC rule
---
.gitignore | 1 +
Snakefile | 5 +++-
TODO | 3 ++-
configuration/define_resources.yaml | 5 ++++
envs/LONGQC.yaml | 13 ++++++++++
rules/build_hifi.smk | 37 ++++++++++++++++++++++++++++-
6 files changed, 61 insertions(+), 3 deletions(-)
create mode 100644 envs/LONGQC.yaml
diff --git a/.gitignore b/.gitignore
index a8bfcdc..5c567d6 100644
--- a/.gitignore
+++ b/.gitignore
@@ -4,6 +4,7 @@
buscoLineage/
buscoLineage
+tools/LongQC/
#ignoring the new logs
slurm_logs/
diff --git a/Snakefile b/Snakefile
index 288ad8d..8facab1 100644
--- a/Snakefile
+++ b/Snakefile
@@ -138,8 +138,11 @@ elif set(['sample', 'hifi_reads', 'meryl_kmer_size','trimSMRTbell', 'fastQC']).i
samples=samples.set_index(['sample','readCounter'])
- ruleAllQCFiles=[]
+ longqc_dir = "tools/LongQC"
+
+ ruleAllQCFiles=[]
+ ### TODO change this column in a generic QC analysis. (both nanoplot and and longqc running)
if samples['fastQC'].str.contains('True').any():
ruleAllQCFiles=[expand(os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/QC/multiqc/{sample}.multiqcReport.html"), sample=key) for key, [value1, value2] in testDictQC.items()]
ruleAll=[expand(os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/merylDb/complete_hifi.{sample}.{kmer}.meryl"), sample=key, kmer=value1) for key, [value1, value2] in dictSamples.items()]
diff --git a/TODO b/TODO
index 23abf39..95f1314 100644
--- a/TODO
+++ b/TODO
@@ -1,4 +1,5 @@
- add nanoplot DONE
- remove cutadapter
- change hifi to long reads or something similar
-- add longqc (optional)
+- add longqc DONE
+- add a column for pacbio/ont
diff --git a/configuration/define_resources.yaml b/configuration/define_resources.yaml
index b70c4dd..4383973 100644
--- a/configuration/define_resources.yaml
+++ b/configuration/define_resources.yaml
@@ -167,6 +167,11 @@ nanoplot_hifi:
time: "12:00:00"
threads: 2
+longQC_hifi:
+ mem_mb: 12000
+ time: "12:00:00"
+ threads: 4
+
multiQC_hifi:
mem_mb: 4000
time: "06:00:00"
diff --git a/envs/LONGQC.yaml b/envs/LONGQC.yaml
new file mode 100644
index 0000000..170b1d9
--- /dev/null
+++ b/envs/LONGQC.yaml
@@ -0,0 +1,13 @@
+channels:
+ - anaconda
+ - conda-forge
+ - bioconda
+dependencies:
+ - matplotlib=3.10.0
+ - scikit-learn=1.6.1
+ - pandas=2.2.3
+ - jinja2=3.1.5
+ - h5py=3.12.1
+ - pysam=0.23.0
+ - edlib=1.2.3
+ - python-edlib=1.3.9.post1
\ No newline at end of file
diff --git a/rules/build_hifi.smk b/rules/build_hifi.smk
index c05f060..2b1bd5f 100644
--- a/rules/build_hifi.smk
+++ b/rules/build_hifi.smk
@@ -113,7 +113,8 @@ rule fastQC_hifi:
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/nanoplot/NanoPlot-report.html"),
+ os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/QC/longQC/results/web_summary.html")
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/"),
@@ -199,3 +200,37 @@ rule NanoPlot:
"""
(NanoPlot -t {threads} -o {params.folder2out} --fastq {input} --plots kde dot) &> {log}
"""
+
+rule longQC:
+ input:
+ 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/")
+ output:
+ os.path.join(config['Results'],"0_buildDatabases/{sample}/hifiReads/QC/longQC/results/web_summary.html")
+ threads:
+ resource['longQC_hifi']['threads']
+ resources:
+ mem_mb=resource['longQC_hifi']['mem_mb'],
+ time=resource['longQC_hifi']['time'],
+ log:
+ os.path.join(config['Results'], "0_buildDatabases/{sample}/hifiReads/logs/longQC.log")
+ conda:
+ os.path.join(workflow.basedir, "envs/LONGQC.yaml")
+ shell:
+ """
+ ## TODO making either pacbio or nanopore for -x
+ (python tools/LongQC/longQC.py sampleqc -x ont-rapid -o {params.folder2out}/results {input.fastq}) &> {log}
+ """
+
+rule install_longqc:
+ output:
+ os.path.join(longqc_dir,"longQC.py")
+ shell:
+ """
+ mkdir -p tools
+ cd tools
+ git clone https://github.com/yfukasawa/LongQC.git
+ cd LongQC/minimap2-coverage && make
+ """
\ No newline at end of file
--
GitLab