seq.yaml

# Full hg38/GRCh38 reference genome distributed by 1000 genomes
# Derived from NCBI set with HLA and decoy alternative alleles
# ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/technical/reference/GRCh38_reference_genome/
---
attributes:
  name: seq
  version: 1000g-20150219_1
recipe:
  full:
    recipe_type: bash
    recipe_cmds:
      - |
        url=ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/technical/reference/GRCh38_reference_genome
        base=GRCh38_full_analysis_set_plus_decoy_hla
        new=hg38
        mkdir -p seq
        for suffix in .dict .fa.fai
        do
          [[ -f seq/$new$suffix ]] || wget -c -O seq/$new$suffix $url/$base$suffix
        done
        # gzipped references for use with ensembl-vep HGVS
        urlgz=https://s3.amazonaws.com/biodata/hg38_bundle
        for suffix in .fa.gz .fa.gz.fai .fa.gz.gzi
        do
          [[ -f seq/$new$suffix ]] || wget --no-check-certificate -c -O seq/$new$suffix $urlgz/$new$suffix
        done
        gunzip -c seq/hg38.fa.gz > seq/hg38.fa
        touch seq/hg38.fa.fai
        touch seq/hg38.dict
    recipe_outfiles:
      - seq/hg38.fa
      - seq/hg38.fa.fai
      - seq/hg38.fa.gz
      - seq/hg38.fa.gz.fai
      - seq/hg38.fa.gz.gzi
      - seq/hg38.dict