annotate test-data/create_test-data @ 0:6e75a84e9338 draft

planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty

#!/bin/bash
source ~/env/bin/activate

samtools faidx chrMT.fa
#TEST1: single read, with everything default
python2 ../genReads.py -r chrMT.fa -R 101 -o out --rng 1
mv out_read1.fq chrMT_read1.fq

#TEST2: PE reads, with everything default
python2 ../genReads.py -r chrMT.fa -R 101 -o out --rng 1 --pe 300 30
mv out_read1.fq chrMT-PE_read1.fq
mv out_read2.fq chrMT-PE_read2.fq

#TEST3: PE reads, with everything default, now with VCF and BAM files
python2 ../genReads.py -r chrMT.fa -R 101 -o out --rng 1 --pe 300 30 --vcf --bam
mv out_read1.fq chrMT-PE-VCF-BAM_read1.fq
mv out_read2.fq chrMT-PE-VCF-BAM_read2.fq
mv out_golden.bam chrMT-PE-VCF-BAM.bam
mv out_golden.vcf chrMT-PE-VCF-BAM.vcf

samtools index chrMT-PE-VCF-BAM.bam

#TEST4: PE reads, with VCF and BAM files and VCF file from TEST3 as the seed
python2 ../genReads.py -r chrMT.fa -R 101 -o out --rng 1 --pe 300 30 --vcf --bam -v chrMT-PE-VCF-BAM.vcf
mv out_read1.fq chrMT-PE-VCF-BAM-vcf_read1.fq
mv out_read2.fq chrMT-PE-VCF-BAM-vcf_read2.fq
mv out_golden.bam chrMT-PE-VCF-BAM-vcf.bam
mv out_golden.vcf chrMT-PE-VCF-BAM-vcf.vcf

samtools index chrMT-PE-VCF-BAM-vcf.bam

#TEST5: PE reads, with VCF and BAM files and BED file as the targeted region
python2 ../genReads.py -r chrMT.fa -R 101 -o out --rng 1 --pe 300 30 --vcf --bam -t chrMT-Targets.bed -to 0.02
mv out_read1.fq chrMT-PE-VCF-BAM-Targeted_read1.fq
mv out_read2.fq chrMT-PE-VCF-BAM-Targeted_read2.fq
mv out_golden.bam chrMT-PE-VCF-BAM-Targeted.bam
mv out_golden.vcf chrMT-PE-VCF-BAM-Targeted.vcf

samtools index chrMT-PE-VCF-BAM-Targeted.bam

#TEST6: PE reads, with everything default, now with VCF and BAM files GZIPPED
python2 ../genReads.py -r chrMT.fa -R 101 -o out --rng 1 --pe 300 30 --vcf --bam --gz
mv out_read1.fq.gz chrMT-PE-VCF-BAM-gz_read1.fq.gz
mv out_read2.fq.gz chrMT-PE-VCF-BAM-gz_read2.fq.gz
mv out_golden.bam chrMT-PE-VCF-BAM-gz.bam
#Galaxy does not support gzipped VCF file (not sure if this is BGZIPPED)
gunzip out_golden.vcf.gz && mv out_golden.vcf chrMT-PE-VCF-BAM-gz.vcf

samtools index chrMT-PE-VCF-BAM-gz.bam

#TEST7: PE reads, with all error parameters changed, with VCF and BAM files, not compressed 
python2 ../genReads.py -r chrMT.fa -R 151 -o out --rng 123 --pe 500 50 --vcf --bam -c 20 -E 0.123 -M 0.123 -p 3
mv out_read1.fq chrMT-PE-VCF-BAM-panic_read1.fq
mv out_read2.fq chrMT-PE-VCF-BAM-panic_read2.fq
mv out_golden.bam chrMT-PE-VCF-BAM-panic.bam
mv out_golden.vcf chrMT-PE-VCF-BAM-panic.vcf

samtools index chrMT-PE-VCF-BAM-panic.bam

#####
# computeGC TESTS
#

#TEST1: Use BAM from TEST3 to create the model file. Window size = 10
bedtools genomecov -d -ibam chrMT-PE-VCF-BAM.bam -g chrMT.fa > chrMT-PE-VCF-BAM.genomecov
python2 ../utilities/computeGC.py -r chrMT.fa -i chrMT-PE-VCF-BAM.genomecov -w 10 -o chrMT-PE-VCF-BAM-computeGC.p

#####
# computeFraglen
#
samtools view chrMT-PE-VCF-BAM.bam | python2 ../utilities/computeFraglen.py
mv fraglen.p chrMT-PE-VCF-BAM-fraglen.p

#####
# genMutModel
#

#TEST1: Default settings
python2 ../utilities/genMutModel.py -r chrMT.fa -m chrMT-PE-VCF-BAM.vcf -o chrMT-PE-VCF-BAM-genMutModel.p

#TEST2: Defined include list
python2 ../utilities/genMutModel.py -r chrMT.fa -m chrMT-PE-VCF-BAM.vcf -o chrMT-PE-VCF-BAM-INCLUDELIST-genMutModel.p\
        -bi chrMT-Targets.bed

#TEST3: Defined exclude list
python2 ../utilities/genMutModel.py -r chrMT.fa -m chrMT-PE-VCF-BAM.vcf -o chrMT-PE-VCF-BAM-EXCLUDELIST-genMutModel.p\
        -be chrMT-Targets.bed
		
#TEST1: Default settings with all booleans set to YES
python2 ../utilities/genMutModel.py -r chrMT.fa -m chrMT-PE-VCF-BAM.vcf -o chrMT-PE-VCF-BAM-BOOLEANS-genMutModel.p\
        --save-trinuc --no-whitelist --skip-common

#####
# genSeqErrorModel
#

#TEST1 - 100,000 simulations - Single read
python2 ../utilities/genSeqErrorModel.py -i chrMT_read1.fq -s 100000 -o chrMT_read1_genSeqErrorModel.p

#TEST2 - 100,000 simulations - Paired reads
python2 ../utilities/genSeqErrorModel.py -i chrMT-PE_read1.fq -i2 chrMT-PE_read2.fq -s 100000 -o chrMT-PE_read1_genSeqErrorModel.p

#TEST3 - 100,000 simulations - Only 100 reads
python2 ../utilities/genSeqErrorModel.py -i chrMT-PE_read1.fq -i2 chrMT-PE_read2.fq -s 100000 -o chrMT-PE-100reads_read1_genSeqErrorModel.p -n 100