test for ggd recipe

bcbio.pbs

@@ -32,7 +32,7 @@ then
 fi
 
 # testing environment of bcbio
-if [ -n $testing ] && [ $testing == "T" ]
+if [ -n $testing ] && [ "$testing" == "T" ]
 then
     . /hpf/largeprojects/ccmbio/naumenko/tools/bcbio_testing/.test_profile
     echo "Using test bcbio instance"

cre.bcbio.upgrade.sh

@@ -63,11 +63,11 @@ which bcbio_nextgen.py
 # --genomes hg38
 
 # gnomad 14h
-# bcbio_nextgen.py upgrade -u skip --genomes GRCh37 --datatarget gnomad
+bcbio_nextgen.py upgrade -u skip --genomes GRCh37 --datatarget gnomad
 # bcbio_nextgen.py upgrade -u skip --genomes hg38 --datatarget gnomad
 
 # dbnsfp
-bcbio_nextgen.py upgrade -u skip --genomes GRCh37 --datatarget dbnsfp
+# bcbio_nextgen.py upgrade -u skip --genomes GRCh37 --datatarget dbnsfp
 
 # rnaseq
 # bcbio_nextgen.py upgrade -u skip --genomes GRCh37 --datatarget rnaseq

cre.cloudbiolinux.prepare_gnomad_test.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+# prepares full chomosome test set for gnomad to test ggd recipe
+prefix=gnomad.exomes.r2.1.sites.grch38.chr
+for chrom in $(seq 1 22;echo X Y)
+do
+  curl -r 0-1000000 -O http://ftp.ensemblorg.ebi.ac.uk/pub/data_files/homo_sapiens/GRCh38/variation_genotype/gnomad/r2.1/exomes/${prefix}${chrom}_noVEP.vcf.gz
+  gunzip -c ${prefix}${chrom}_noVEP.vcf.gz | head -n 1400 > ${prefix}${chrom}_noVEP.vcf
+  bgzip -f ${prefix}${chrom}_noVEP.vcf
+  tabix ${prefix}${chrom}_noVEP.vcf.gz
+done

cre.rohet.naive.sh

@@ -19,8 +19,9 @@ then
     maf=$3
 fi
 #depth is for 3 samples
-gemini query -q "select chrom,start+1 as pos, ref, alt,impact,qual,depth, gene, max_aaf_all as maf, gts."$sample",gt_types."$sample",gt_alt_depths."$sample" from variants where 
-type='snp' and depth>=10 and qual>=500 and max_aaf_all<="$maf --gt-filter "gt_types."$sample" != 2" $2 | grep -v chrGL | sed s/chr// | sed s/"\t"/","/g | sort -t "," -k1,1 -k2,2n \
+# depth is in old databases
+gemini query -q "select chrom,start+1 as pos, ref, alt,impact,qual,dp, gene, max_aaf_all as maf, gts."$sample",gt_types."$sample",gt_alt_depths."$sample" from variants where 
+type='snp' and dp>=10 and qual>=500 and max_aaf_all<="$maf --gt-filter "gt_types."$sample" != 2" $2 | grep -v chrGL | sed s/chr// | sed s/"\t"/","/g | sort -t "," -k1,1 -k2,2n \
 | tee -a $sample.rohet_variants.csv | awk -F "," '
 BEGIN{
     prev_genotype=0;