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

planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty

1 #!/bin/bash
2 source ~/env/bin/activate
3 
4 samtools faidx chrMT.fa
5 #TEST1: single read, with everything default
6 python2 ../genReads.py -r chrMT.fa -R 101 -o out --rng 1
7 mv out_read1.fq chrMT_read1.fq
8 
9 #TEST2: PE reads, with everything default
10 python2 ../genReads.py -r chrMT.fa -R 101 -o out --rng 1 --pe 300 30
11 mv out_read1.fq chrMT-PE_read1.fq
12 mv out_read2.fq chrMT-PE_read2.fq
13 
14 #TEST3: PE reads, with everything default, now with VCF and BAM files
15 python2 ../genReads.py -r chrMT.fa -R 101 -o out --rng 1 --pe 300 30 --vcf --bam
16 mv out_read1.fq chrMT-PE-VCF-BAM_read1.fq
17 mv out_read2.fq chrMT-PE-VCF-BAM_read2.fq
18 mv out_golden.bam chrMT-PE-VCF-BAM.bam
19 mv out_golden.vcf chrMT-PE-VCF-BAM.vcf
20 
21 samtools index chrMT-PE-VCF-BAM.bam
22 
23 #TEST4: PE reads, with VCF and BAM files and VCF file from TEST3 as the seed
24 python2 ../genReads.py -r chrMT.fa -R 101 -o out --rng 1 --pe 300 30 --vcf --bam -v chrMT-PE-VCF-BAM.vcf
25 mv out_read1.fq chrMT-PE-VCF-BAM-vcf_read1.fq
26 mv out_read2.fq chrMT-PE-VCF-BAM-vcf_read2.fq
27 mv out_golden.bam chrMT-PE-VCF-BAM-vcf.bam
28 mv out_golden.vcf chrMT-PE-VCF-BAM-vcf.vcf
29 
30 samtools index chrMT-PE-VCF-BAM-vcf.bam
31 
32 #TEST5: PE reads, with VCF and BAM files and BED file as the targeted region
33 python2 ../genReads.py -r chrMT.fa -R 101 -o out --rng 1 --pe 300 30 --vcf --bam -t chrMT-Targets.bed -to 0.02
34 mv out_read1.fq chrMT-PE-VCF-BAM-Targeted_read1.fq
35 mv out_read2.fq chrMT-PE-VCF-BAM-Targeted_read2.fq
36 mv out_golden.bam chrMT-PE-VCF-BAM-Targeted.bam
37 mv out_golden.vcf chrMT-PE-VCF-BAM-Targeted.vcf
38 
39 samtools index chrMT-PE-VCF-BAM-Targeted.bam
40 
41 #TEST6: PE reads, with everything default, now with VCF and BAM files GZIPPED
42 python2 ../genReads.py -r chrMT.fa -R 101 -o out --rng 1 --pe 300 30 --vcf --bam --gz
43 mv out_read1.fq.gz chrMT-PE-VCF-BAM-gz_read1.fq.gz
44 mv out_read2.fq.gz chrMT-PE-VCF-BAM-gz_read2.fq.gz
45 mv out_golden.bam chrMT-PE-VCF-BAM-gz.bam
46 #Galaxy does not support gzipped VCF file (not sure if this is BGZIPPED)
47 gunzip out_golden.vcf.gz && mv out_golden.vcf chrMT-PE-VCF-BAM-gz.vcf
48 
49 samtools index chrMT-PE-VCF-BAM-gz.bam
50 
51 #TEST7: PE reads, with all error parameters changed, with VCF and BAM files, not compressed 
52 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
53 mv out_read1.fq chrMT-PE-VCF-BAM-panic_read1.fq
54 mv out_read2.fq chrMT-PE-VCF-BAM-panic_read2.fq
55 mv out_golden.bam chrMT-PE-VCF-BAM-panic.bam
56 mv out_golden.vcf chrMT-PE-VCF-BAM-panic.vcf
57 
58 samtools index chrMT-PE-VCF-BAM-panic.bam
59 
60 #####
61 # computeGC TESTS
62 #
63 
64 #TEST1: Use BAM from TEST3 to create the model file. Window size = 10
65 bedtools genomecov -d -ibam chrMT-PE-VCF-BAM.bam -g chrMT.fa > chrMT-PE-VCF-BAM.genomecov
66 python2 ../utilities/computeGC.py -r chrMT.fa -i chrMT-PE-VCF-BAM.genomecov -w 10 -o chrMT-PE-VCF-BAM-computeGC.p
67 
68 #####
69 # computeFraglen
70 #
71 samtools view chrMT-PE-VCF-BAM.bam | python2 ../utilities/computeFraglen.py
72 mv fraglen.p chrMT-PE-VCF-BAM-fraglen.p
73 
74 #####
75 # genMutModel
76 #
77 
78 #TEST1: Default settings
79 python2 ../utilities/genMutModel.py -r chrMT.fa -m chrMT-PE-VCF-BAM.vcf -o chrMT-PE-VCF-BAM-genMutModel.p
80 
81 #TEST2: Defined include list
82 python2 ../utilities/genMutModel.py -r chrMT.fa -m chrMT-PE-VCF-BAM.vcf -o chrMT-PE-VCF-BAM-INCLUDELIST-genMutModel.p\
83         -bi chrMT-Targets.bed
84 
85 #TEST3: Defined exclude list
86 python2 ../utilities/genMutModel.py -r chrMT.fa -m chrMT-PE-VCF-BAM.vcf -o chrMT-PE-VCF-BAM-EXCLUDELIST-genMutModel.p\
87         -be chrMT-Targets.bed
88 		
89 #TEST1: Default settings with all booleans set to YES
90 python2 ../utilities/genMutModel.py -r chrMT.fa -m chrMT-PE-VCF-BAM.vcf -o chrMT-PE-VCF-BAM-BOOLEANS-genMutModel.p\
91         --save-trinuc --no-whitelist --skip-common
92 
93 #####
94 # genSeqErrorModel
95 #
96 
97 #TEST1 - 100,000 simulations - Single read
98 python2 ../utilities/genSeqErrorModel.py -i chrMT_read1.fq -s 100000 -o chrMT_read1_genSeqErrorModel.p
99 
00 #TEST2 - 100,000 simulations - Paired reads
01 python2 ../utilities/genSeqErrorModel.py -i chrMT-PE_read1.fq -i2 chrMT-PE_read2.fq -s 100000 -o chrMT-PE_read1_genSeqErrorModel.p
02 
03 #TEST3 - 100,000 simulations - Only 100 reads
04 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