annotate test-data/generate_tests.sh @ 2:396cf2c12173 draft

"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit 186b72f369eb2a11d92f4d63cac2e8ebe386b9bd"
author iuc
date Mon, 13 Dec 2021 15:15:14 +0000
parents 8ebb6520eef4
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
1 #! /usr/bin/bash
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
2 ## Generate input data:
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
3 if [ ! -e test-data/small_simplified.gtf ]; then
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
4 wget http://ftp.ensembl.org/pub/release-102/gtf/mus_musculus/Mus_musculus.GRCm38.102.gtf.gz -O /tmp/Mus_musculus.GRCm38.102.gtf.gz
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
5 zcat /tmp/Mus_musculus.GRCm38.102.gtf.gz | head -n 5 > test-data/small.gtf
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
6 zcat /tmp/Mus_musculus.GRCm38.102.gtf.gz | awk -v OFS="\t" -v start=74667792 -v end=74748393 '$1 == "2" && $5 > start && $4 < end{print "chr"$0}' >> test-data/small.gtf
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
7 # annotatePeaks.pl gives different results all time. I need to simplify the gtf.
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
8 cat test-data/small.gtf | grep -v -P "ENSMUST00000152027|ENSMUST00000156342|ENSMUST00000139005|ENSMUST00000144544|ENSMUST00000111982|ENSMUST00000140666|ENSMUST00000190553|ENSMUST00000132326|ENSMUST00000047830|ENSMUST00000047904|ENSMUST00000111980|ENSMUSG00000065500|ENSMUSG00000100642" > test-data/small_simplified.gtf
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
9 fi
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
10 if [ ! -e test-data/CTCF_peaks.bed ]; then
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
11 wget https://raw.githubusercontent.com/lldelisle/scriptsForWilleminEtAl2021/main/CTCF/E12_Limbs_Wt_CTCF_colored.bed -O test-data/CTCF_peaks.bed
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
12 fi
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
13 if [ ! -e test-data/CTCF_peaks_shifted.bed ]; then
2
396cf2c12173 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit 186b72f369eb2a11d92f4d63cac2e8ebe386b9bd"
iuc
parents: 0
diff changeset
14 cat test-data/CTCF_peaks.bed | grep "chr2" | awk -v OFS="\t" '$3<75000000 && $2>73740000{$1="mm10_dna"; $2-=73740000; $3-=73740000; print}' > test-data/CTCF_peaks_shifted.bed
0
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
15 fi
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
16 # chr2_subset.fa was downloaded from UCSC
2
396cf2c12173 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit 186b72f369eb2a11d92f4d63cac2e8ebe386b9bd"
iuc
parents: 0
diff changeset
17 # https://genome.ucsc.edu/cgi-bin/hgc?hgsid=1234982067_JnS4z30UVCNarTg26Ztd1Oh6nfu6&g=htcGetDna2&table=&i=mixed&o=56694975&l=56694975&r=56714605&getDnaPos=chr2%3A73740000-75000000&db=mm10&hgSeq.cdsExon=1&hgSeq.padding5=0&hgSeq.padding3=0&hgSeq.casing=upper&hgSeq.maskRepeats=on&boolshad.hgSeq.maskRepeats=0&hgSeq.repMasking=lower&boolshad.hgSeq.revComp=0&submit=get+DNA
0
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
18
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
19 . <(planemo conda_env homer_gtf_to_annotation.xml)
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
20 echo "$(which homer)"
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
21 ## homer_gtf_to_annotation
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
22 ## First test
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
23 parseGTF.pl test-data/small.gtf ann -features exon start_codon stop_codon > /tmp/annotations.txt
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
24 assignGenomeAnnotation /tmp/annotations.txt /tmp/annotations.txt -prioritize test-data/annotations_default.txt > test-data/annotations_default_stats.txt
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
25 ## Second test
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
26 parseGTF.pl test-data/small.gtf ann -features exon start_codon > /tmp/annotations2.txt
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
27 assignGenomeAnnotation /tmp/annotations2.txt /tmp/annotations2.txt -prioritize test-data/annotations_exon_start.txt > test-data/annotations_exon_start_stats.txt
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
28 ## Third test
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
29 parseGTF.pl test-data/small.gtf ann -features exon start_codon stop_codon -annTSSstartOffset -50 -annTSSendOffset 50 -annTTSstartOffset -50 -annTTSendOffset 50 > /tmp/annotations3.txt
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
30 assignGenomeAnnotation /tmp/annotations3.txt /tmp/annotations3.txt -prioritize test-data/annotations_small_TSSTTS.txt > test-data/annotations_small_TSSTTS_stats.txt
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
31
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
32 ## For annotatePeaks.pl
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
33 parseGTF.pl test-data/small_simplified.gtf ann -features exon start_codon stop_codon > /tmp/annotations.txt
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
34 assignGenomeAnnotation /tmp/annotations.txt /tmp/annotations.txt -prioritize test-data/annotations_default_simplified.txt
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
35
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
36 ## homer_annotatePeaks
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
37 ## First test
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
38 annotatePeaks.pl test-data/CTCF_peaks.bed none -gtf test-data/small_simplified.gtf -ann test-data/annotations_default_simplified.txt > test-data/CTCF_peaks_first.txt
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
39 ## Second test
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
40 annotatePeaks.pl test-data/CTCF_peaks.bed none -ann test-data/annotations_default.txt > test-data/CTCF_peaks_second.txt
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
41 ## Third test
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
42 annotatePeaks.pl test-data/CTCF_peaks.bed none -gtf test-data/small_simplified.gtf > test-data/CTCF_peaks_third.txt
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
43 ## Fourth test
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
44 annotatePeaks.pl test-data/fake_phix_peaks.bed test-data/phiX174.fasta -CpG > test-data/phiXcpg.txt
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
45 ## Fifth test
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
46 annotatePeaks.pl test-data/fake_phix_peaks.bed none > test-data/phiX_nothing.txt
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
47
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
48 ## findMotifsGenome
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
49 # ! Genome preparsing is giving different results...
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
50 findMotifsGenome.pl test-data/fake_phix_peaks.bed test-data/phiX174.fasta fake_phix_peaks_bed_motif
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
51 mv fake_phix_peaks_bed_motif test-data/motif_test1
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
52 # Thus I needed to use has_text for the other outputs
2
396cf2c12173 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit 186b72f369eb2a11d92f4d63cac2e8ebe386b9bd"
iuc
parents: 0
diff changeset
53 # gunzip -c test-data/chr2_subset.fa.gz > test-data/chr2_subset.fa
0
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
54 # findMotifsGenome.pl test-data/CTCF_peaks_shifted.bed test-data/chr2_subset.fa CTCF_peaks_shifted_bed_motif
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
55 # mv CTCF_peaks_shifted_bed_motif test-data/motif_test2
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
56 # findMotifsGenome.pl test-data/CTCF_peaks_shifted.bed test-data/chr2_subset.fa CTCF_peaks_shifted_bed_motif -mask
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
57 # mv CTCF_peaks_shifted_bed_motif test-data/motif_test3
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
58 # findMotifsGenome.pl test-data/CTCF_peaks_shifted.bed test-data/chr2_subset.fa CTCF_peaks_shifted_bed_motif -mset plants -nomotif
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
59 # mv CTCF_peaks_shifted_bed_motif test-data/motif_test4
8ebb6520eef4 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/homer commit e49d856e0181edc6120220a1b819cba2466a4289"
iuc
parents:
diff changeset
60