bug fix

naumenko-sa · Aug 24, 2018 · 7a1398c · 7a1398c
1 parent 86bbe2c
commit 7a1398c
Show file tree

Hide file tree

Showing 3 changed files with 9 additions and 34 deletions.
diff --git a/cram2fq.sh b/cram2fq.sh
@@ -22,4 +22,4 @@ cramtools fastq -Xmx10g -F $sample --skip-md5-check \
 		-R /hpf/largeprojects/ccmbio/naumenko/tools/bcbio/genomes/Hsapiens/hg19/seq/hg19.fa
 
 mv $sample_1.fastq.gz $sample_1.fq.gz
-mv $sample_2.fastq.gz $sample_2.fq.gz
+mv $sample_2.fastq.gz $sample_2.fq.gz
diff --git a/cre.gemini2txt.vcf2db.sh b/cre.gemini2txt.vcf2db.sh
@@ -56,9 +56,9 @@ sQuery="select \
         start+1 as Pos,\
         aa_change as AA_change,\
         hgvsc as Codon_change,\
-        af_esp_aa as Evs_af_aa,\
-        af_esp_ea as Evs_af_ea,\
-        af_esp_all as Evs_af_all,\
+        af_evs_aa as Evs_af_aa,\
+        af_evs_ea as Evs_af_ea,\
+        af_evs as Evs_af_all,\
         phylop20way_mammalian as Conserved_in_20_mammals,
         gts,"
 

diff --git a/cre.vep.sh b/cre.vep.sh
@@ -1,9 +1,9 @@
 #/bin/bash
 
-# gemini.vcf2vep - annotate vcf with vep before loading to gemini database
+# gemini.vcf2vep annotates vcf with vep before loading to gemini database
 # based on  bcbio.log
-# uses hgvs notation and no --pick - all effects for a gene
-# 10h is not enough for genomes or big multisample vcfs
+# uses hgvs notation and no --pick = all effects for a gene
+# 10h walltime is not enough for genomes or big multisample vcfs
 
 #PBS -l walltime=23:00:00,nodes=1:ppn=1
 #PBS -joe .
@@ -17,32 +17,7 @@ fi
 
 bname=`basename $vcf .vcf.gz`
 
-# old one for  VEP87
-#unset PERL5LIB && export PATH=/hpf/largeprojects/ccmbio/naumenko/tools/bcbio/anaconda/bin:$PATH && /home/naumenko/work/tools/bcbio/anaconda/bin/variant_effect_predictor.pl --vcf -o stdout \
-#    -i $vcf --species homo_sapiens --no_stats --cache --offline --dir /hpf/largeprojects/ccmbio/naumenko/tools/bcbio/genomes/Hsapiens/GRCh37/vep --symbol --numbers --biotype --total_length \
-#    --canonical --gene_phenotype --ccds --fields Consequence,Codons,Amino_acids,Gene,SYMBOL,Feature,EXON,PolyPhen,SIFT,Protein_position,BIOTYPE,CANONICAL,CCDS,HGVSc,HGVSp \
-#    --plugin LoF,human_ancestor_fa:/hpf/largeprojects/ccmbio/naumenko/tools/bcbio/genomes/Hsapiens/GRCh37/variation/human_ancestor.fa.gz --sift b --polyphen b --hgvs --shift_hgvs 1 \
-#    --fasta /hpf/largeprojects/ccmbio/naumenko/tools/bcbio/genomes/Hsapiens/GRCh37/seq/GRCh37.fa \
-#    | sed '/^#/! s/;;/;/g' | bgzip -c > $bname.vepeffects.vcf.gz
-
-# old one for VEP88
-#unset PERL5LIB && export PATH=/hpf/largeprojects/ccmbio/naumenko/tools/bcbio/anaconda/bin:$PATH && /hpf/largeprojects/ccmbio/naumenko/tools/bcbio/anaconda/bin/vep \
-#    --vcf -o stdout -i $vcf --fork 7 --species homo_sapiens --no_stats --cache --offline --dir /hpf/largeprojects/ccmbio/naumenko/tools/bcbio/genomes/Hsapiens/GRCh37/vep \
-#    --symbol --numbers --biotype --total_length --canonical --gene_phenotype --ccds --uniprot --domains --regulatory --protein --tsl --appris --af --max_af --af_1kg --af_esp \
-#    --af_exac --pubmed --variant_class --fasta /hpf/largeprojects/ccmbio/naumenko/tools/bcbio/genomes/Hsapiens/GRCh37/seq/GRCh37.fa.gz \
-#    --plugin LoF,human_ancestor_fa:/hpf/largeprojects/ccmbio/naumenko/tools/bcbio/genomes/Hsapiens/GRCh37/variation/human_ancestor.fa.gz --sift b --polyphen b --hgvs --shift_hgvs 1 --merged \
-#    | sed '/^#/! s/;;/;/g' | bgzip -c > $bname.vepeffects.vcf.gz
-
-#old one for VEP90
-#unset PERL5LIB && export PATH=/hpf/largeprojects/ccmbio/naumenko/tools/bcbio/anaconda/bin:$PATH && /hpf/largeprojects/ccmbio/naumenko/tools/bcbio/anaconda/bin/vep --vcf -o stdout \
-#    -i $vcf --fork 5 --species homo_sapiens --no_stats --cache --offline --dir /hpf/largeprojects/ccmbio/naumenko/tools/bcbio/genomes/Hsapiens/GRCh37/vep --symbol --numbers --biotype --total_length \
-#    --canonical --gene_phenotype --ccds --uniprot --domains --regulatory --protein --tsl --appris --af --max_af --af_1kg --af_esp --af_gnomad --pubmed --variant_class \
-#    --fasta /hpf/largeprojects/ccmbio/naumenko/tools/bcbio/genomes/Hsapiens/GRCh37/seq/GRCh37.fa.gz \
-#    --plugin LoF,human_ancestor_fa:/hpf/largeprojects/ccmbio/naumenko/tools/bcbio/genomes/Hsapiens/GRCh37/variation/human_ancestor.fa.gz \
-#    --plugin MaxEntScan,/hpf/largeprojects/ccmbio/naumenko/tools/bcbio/anaconda/share/maxentscan-0_2004.04.21-0 --plugin SpliceRegion --sift b --polyphen b --hgvs --shift_hgvs 1 --merged \
-#    | sed '/^#/! s/;;/;/g' | bgzip -c > $bname.vepeffects.vcf.gz
-
-#the old one
+#VEP91.2
 #unset PERL5LIB && export PATH=/hpf/largeprojects/ccmbio/naumenko/tools/bcbio/anaconda/bin:$PATH && /hpf/largeprojects/ccmbio/naumenko/tools/bcbio/anaconda/bin/vep --vcf -o stdout \
 #    -i $vcf --fork 5 --species homo_sapiens --no_stats --cache --offline --dir /hpf/largeprojects/ccmbio/naumenko/tools/bcbio/genomes/Hsapiens/GRCh37/vep --symbol --numbers --biotype --total_length \
 #    --canonical --gene_phenotype --ccds --uniprot --domains --regulatory --protein --tsl --appris --af --max_af --af_1kg --af_esp --af_gnomad --pubmed --variant_class \
@@ -68,4 +43,4 @@ unset PERL5LIB && vep --vcf -o stdout \
     --plugin SpliceRegion --sift b --polyphen b --hgvs --shift_hgvs 1 --merged \
     | sed '/^#/! s/;;/;/g' | bgzip -c > $bname.vepeffects.vcf.gz
 
-tabix $bname.vepeffects.vcf.gz
+tabix $bname.vepeffects.vcf.gz