comparison test-data/util/build-gemini-testdata.sh @ 4:cdd90678004a draft

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/gemini commit 283362494058ed64143b1f27afb447b8a1cb4313
author iuc
date Fri, 14 Dec 2018 12:47:19 -0500
parents
children
comparison
equal deleted inserted replaced
3:ee894347fcd6 4:cdd90678004a
1 cd "$(dirname "$0")"
2
3 export GEMINI_CONFIG=../test-cache
4 OUT_PTH=$GEMINI_CONFIG/gemini/data
5 GENOMIC_REGION=3:187000000-187500000
6
7
8 if [ -n "$1" ]; then
9
10 IN_PTH="$1"
11 # downsample all vcf and bed annotation files to the region of interest and reindex
12 for vcf in `ls $IN_PTH/*.gz | grep -v hprd_interaction_edges.gz -`
13 do
14 python ./shrink_tabix.py $vcf -r $GENOMIC_REGION -o $OUT_PTH/`basename $vcf`
15 done
16
17 # downsample gene_table files to the region of interest
18 echo "$IN_PTH/summary_gene_table_v75 -> $OUT_PTH/summary_gene_table_v75"
19 python ./shrink_simple_tab.py $IN_PTH/summary_gene_table_v75 -r chr$GENOMIC_REGION -c 0 8 9 -n 1 -o $OUT_PTH/summary_gene_table_v75
20
21 echo "$IN_PTH/detailed_gene_table_v75 -> $OUT_PTH/detailed_gene_table_v75"
22 python ./shrink_simple_tab.py $IN_PTH/detailed_gene_table_v75 -r chr$GENOMIC_REGION -c 0 11 12 -n 1 -o $OUT_PTH/detailed_gene_table_v75
23
24 # filter kegg_pathway files to retain only records of the genes listed
25 # in the downsampled summary_gene_table
26 for kegg in `ls $IN_PTH/kegg_pathways_*`
27 do
28 echo "$kegg -> $OUT_PTH/`basename $kegg`"
29 cut -f2 $OUT_PTH/summary_gene_table_v75 | grep -Fv None | grep -Fwf - $kegg > $OUT_PTH/`basename $kegg`
30 done
31
32 # filter hprd_interaction file to retain only records of the genes listed
33 # in the downsampled summary_gene_table
34 echo "$IN_PTH/hprd_interaction_edges.gz -> $OUT_PTH/hprd_interaction_edges.gz"
35 bgzip -dc $IN_PTH/hprd_interaction_edges.gz > $OUT_PTH/hprd_interaction_edges
36 cut -f2 $OUT_PTH/summary_gene_table_v75 | grep -Fv None | grep -Ff - $OUT_PTH/hprd_interaction_edges | bgzip > $OUT_PTH/hprd_interaction_edges.gz
37 rm $OUT_PTH/hprd_interaction_edges
38
39 # filter cancer_gene_census file to retain only records of the genes listed
40 # in the downsampled summary_gene_table;
41 # TO DO: make the filter stricter by looking for matches only in the first
42 # column of the cancer_gene_census file (but the file is relatively small anyway)
43 echo "$IN_PTH/cancer_gene_census.20140120.tsv -> $OUT_PTH/cancer_gene_census.20140120.tsv"
44 cut -f2 $OUT_PTH/summary_gene_table_v75 | grep -Fv None | grep -Fwf - $IN_PTH/cancer_gene_census.20140120.tsv > $OUT_PTH/cancer_gene_census.20140120.tsv
45
46 else
47 echo "no path to gemini annotation files provided - only building test databases"
48 fi
49
50
51 # now use gemini load to build the test databases
52 echo "Building gemini test databases"
53 echo "Test databases for gemini_load"
54 gemini --annotation-dir $OUT_PTH load --skip-cadd --skip-gerp-bp -v build-data/gemini_load_input.vcf -t snpEff ../gemini_load_result1.db
55 gemini --annotation-dir $OUT_PTH load --skip-cadd --skip-gerp-bp -v build-data/gemini_load_input.vcf -t snpEff --skip-gene-tables --no-load-genotypes ../gemini_load_result2.db
56 echo "Test database for gemini_amend"
57 gemini --annotation-dir $OUT_PTH load --skip-cadd --skip-gerp-bp -v build-data/test.auto_rec.vcf -t snpEff ../gemini_amend_input.db
58 echo "Test database for gemini_annotate"
59 bgzip -c build-data anno.bed > build-data/anno.bed.gz
60 tabix --force -p bed build-data/anno.bed.gz
61 cp ../gemini_load_result1.db ../gemini_annotate_result.db
62 gemini --annotation-dir $OUT_PTH annotate -f build-data/anno.bed.gz -c anno5 -a count ../gemini_annotate_result.db
63 echo "Test database for gemini_set_somatic"
64 cp ../gemini_load_result1.db ../gemini_is_somatic_result.db
65 gemini set_somatic --min-somatic-score 5.65 ../gemini_is_somatic_result.db
66 echo "Test database for gemini_de_novo and gemini_mendel_errors"
67 gemini --annotation-dir $OUT_PTH load --skip-cadd --skip-gerp-bp -v build-data/test.de_novo.vcf -p build-data/test.de_novo.ped -t snpEff ../gemini_de_novo_input.db
68 echo "Test database for gemini_comp_hets"
69 gemini --annotation-dir $OUT_PTH load --skip-cadd --skip-gerp-bp -v build-data/test.comp_het.vcf -p build-data/test.comp_het.ped -t snpEff ../gemini_comphets_input.db
70 echo "Test databases for gemini_autosomal"
71 gemini --annotation-dir $OUT_PTH load --skip-cadd --skip-gerp-bp -v build-data/test.auto_rec.vcf -p build-data/test.auto_rec.ped -t snpEff ../gemini_auto_rec_input.db
72 gemini --annotation-dir $OUT_PTH load --skip-cadd --skip-gerp-bp -v build-data/test.auto_dom.vcf -p build-data/test.auto_dom.ped -t snpEff ../gemini_auto_dom_input.db