dvc.yaml

stages:
  start:
    desc: "convenience stage to start services"
    frozen: true
    cmd: docker-compose up
  stop:
    desc: "convenience stage to stop services"
    frozen: true
    cmd: docker-compose down
  prepare_genome:
    deps:
      - data/ensembl/103/species/Homo_sapiens/Homo_sapiens.GRCh38.dna.primary_assembly.fa.gz
    cmd: >-
      gunzip -k data/ensembl/103/species/Homo_sapiens/Homo_sapiens.GRCh38.dna.primary_assembly.fa.gz &&
      samtools faidx data/ensembl/103/species/Homo_sapiens/Homo_sapiens.GRCh38.dna.primary_assembly.fa &&
      samtools dict data/ensembl/103/species/Homo_sapiens/Homo_sapiens.GRCh38.dna.primary_assembly.fa -o data/ensembl/103/species/Homo_sapiens/Homo_sapiens.GRCh38.dna.primary_assembly.fa.dict
    outs:
      - data/ensembl/103/species/Homo_sapiens/Homo_sapiens.GRCh38.dna.primary_assembly.fa
      - data/ensembl/103/species/Homo_sapiens/Homo_sapiens.GRCh38.dna.primary_assembly.fa.fai
      - data/ensembl/103/species/Homo_sapiens/Homo_sapiens.GRCh38.dna.primary_assembly.fa.dict
  prepare_vep_cache:
    deps:
      - data/ensembl/103/cache/homo_sapiens_vep_103_GRCh38.tar.gz
    cmd: tar -zxf data/ensembl/103/cache/homo_sapiens_vep_103_GRCh38.tar.gz -C data/ensembl/103/cache/
    outs:
      - data/ensembl/103/cache/homo_sapiens
  prepare_vep_plugins:
    deps:
      - data/ensembl/103/103.zip
    cmd: >-
      unzip -u data/ensembl/103/103.zip -d data/ensembl/103/ &&
      mv data/ensembl/103/VEP_plugins-release-103 data/ensembl/103/plugins
    outs:
      - data/ensembl/103/plugins/
  prepare_annotations_digenet:
    deps:
      - data/gwas/annotations/all_variant_disease_pmid_associations.tsv.gz
    cmd: >-
      gunzip -c data/gwas/annotations/all_variant_disease_pmid_associations.tsv.gz |
         awk '($1 ~ /^snpId/ || $2 ~ /NA/) {next} {print $0}' |
         sort -t $'\t' -k2,2 -k3,3n |
         awk '{ gsub (/\t +/, "\t", $0); print}' |
         bgzip -c > data/gwas/annotations/all_variant_disease_pmid_associations_final.tsv.gz
    outs:
      - data/gwas/annotations/all_variant_disease_pmid_associations_final.tsv.gz
  prepare_annotations_clinvar:
    deps:
      - data/gwas/annotations/clinvar.vcf.gz
      - data/gwas/annotations/clinvar.vcf.gz.tbi
    cmd: echo "Clinvars annotations prepared"       
# https://www.ebi.ac.uk/gene2phenotype/downloads/ has no E-Tag!
  prepare_annotations_gene2phenotype: 
    deps:
      - data/gwas/annotations/gene2phenotype/DDG2P.csv
      - data/gwas/annotations/gene2phenotype/CancerG2P.csv
      - data/gwas/annotations/gene2phenotype/EyeG2P.csv
      - data/gwas/annotations/gene2phenotype/SkinG2P.csv
    cmd: >- 
      rm -f data/gwas/annotations/gene2phenotype/AllG2P.csv &&
      awk '(NR == 1) || (FNR > 1)' data/gwas/annotations/gene2phenotype/*.csv 
      > data/gwas/annotations/gene2phenotype/AllG2P.csv
#     && for f in data/gwas/annotations/gene2phenotype/*.gz ; do gunzip -c "$f" > data/gwas/annotations/gene2phenotype/"${f%.*}" ; done 
    outs:
      - data/gwas/annotations/gene2phenotype/AllG2P.csv
  install_opencravat:
    cmd: >-
      oc module install-base && oc module install -y clinvar clinvar_acmg biogrid uk10k_cohort uniprot pubmed provean polyphen2 cardioboost civic civic_gene cosmic cosmic_gene go sift ensembl_regulatory_build dgi cvdkp
  install_cancer:
    cmd: >-
      oc module install chasmplus cgc cgl cancer_genome_interpreter cancer_hotspots civic