Homec4science
Diffusion gtn (master) Browse

Gtnseq/Snakefile_Localhost
510e350ccc65master

Last Change
Raw File

Snakefile_Localhost
View Options

import os
SAMPLES = ['DATASET_NAME']
group = '-'.join(SAMPLES[0].split('-')[1:])
dataset = SAMPLES[0]
# This is for localhost execution
PATH_TO_INPUT_FILES = os.path.join('/input/',group,dataset)
PATH_TO_OUTPUT = os.path.join('/output/',dataset)
for smp in SAMPLES:
print("Sample " + smp + " will be processed")
rule final:
input:
expand(os.path.join(PATH_TO_OUTPUT,'{sample}_R1_fastqc.html'), sample=SAMPLES),
expand(os.path.join(PATH_TO_OUTPUT,'{sample}_R2_fastqc.html'), sample=SAMPLES),
expand(os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed_R1.fastq'), sample=SAMPLES),
expand(os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed_R2.fastq'), sample=SAMPLES),
expand(os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed_R1_fastqc.html'), sample=SAMPLES),
expand(os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed_R2_fastqc.html'), sample=SAMPLES),
expand(os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed.fastq'), sample=SAMPLES),
rule perform_qc:
input:
R1=os.path.join(PATH_TO_INPUT_FILES,'{sample}_R1.fastq'),
R2=os.path.join(PATH_TO_INPUT_FILES,'{sample}_R2.fastq'),
params:
out_dir = PATH_TO_OUTPUT,
output:
os.path.join(PATH_TO_OUTPUT,'{sample}_R1_fastqc.html'),
os.path.join(PATH_TO_OUTPUT,'{sample}_R1_fastqc.zip'),
os.path.join(PATH_TO_OUTPUT,'{sample}_R2_fastqc.html'),
os.path.join(PATH_TO_OUTPUT,'{sample}_R2_fastqc.zip'),
shell:
r'''
fastqc -o {params.out_dir} -f fastq {input.R1} {input.R2}
'''
rule trimmometic_run:
input:
R1_read=os.path.join(PATH_TO_INPUT_FILES,'{sample}_R1.fastq'),
R2_read=os.path.join(PATH_TO_INPUT_FILES,'{sample}_R1.fastq'),
output:
R1_trimmed=os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed_R1.fastq'),
R2_trimmed=os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed_R2.fastq'),
R1_unpaired=os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed_upaired_R1.fastq'),
R2_unpaired=os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed_upaired_R2.fastq'),
message: """---TRIMMOMATIC---"""
shell:
r'''
java -jar /usr/share/java/trimmomatic-0.35.jar PE -phred33 {input.R1_read} {input.R2_read} {output.R1_trimmed} {output.R1_unpaired} {output.R2_trimmed} {output.R2_unpaired} LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 MINLEN:30
'''
rule perform_qc_trimmed:
input:
R1_trimmed_qc=os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed_R1.fastq'),
R2_trimmed_qc=os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed_R2.fastq'),
params:
out_dir=PATH_TO_OUTPUT
output:
os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed_R1_fastqc.html'),
os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed_R1_fastqc.zip'),
os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed_R2_fastqc.html'),
os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed_R2_fastqc.zip'),
message: """---QC trimmed Data---"""
shell:
r'''
fastqc -outdir {params.out_dir} -f fastq {input.R1_trimmed_qc} {input.R2_trimmed_qc}
'''
rule merge_trimmed_R1_and_R2:
input:
R1_trimmed_qc=os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed_R1.fastq'),
R2_trimmed_qc=os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed_R2.fastq'),
output:
merged=os.path.join(PATH_TO_OUTPUT,'{sample}_trimmed.fastq'),
message: """---Merge trimmed R1 and R2---"""
shell:
r'''
cat {input.R1_trimmed_qc} {input.R2_trimmed_qc} > {output.merged}
'''
c4science · Help