unit-testing.Listeria.Kraken1.yml

# This is a subsampling unit test to get early results
on: 
  push:
    branches: [master, dev, esearch-input]
name: Listeria-with-Kraken1

env:
  TSV: "Kalamari/src/Listeria.tsv"
  OUTDIR: "Listeria.out"

jobs:
  build:
    runs-on: ${{ matrix.os }}
    strategy:
      matrix:
        os: ['ubuntu-20.04' ]
        perl: [ '5.32' ]
    name: Perl ${{ matrix.perl }} on ${{ matrix.os }}
    steps:
      - name: Set up perl
        uses: shogo82148/actions-setup-perl@v1
        with:
          perl-version: ${{ matrix.perl }}
          multi-thread: "true"
      - name: checkout my repo
        uses: actions/checkout@v4
        with:
          path: Kalamari

      - name: env check
        run:  |
          echo $PATH | tr ':' '\n' | sort
      - name: apt-get install
        run:  sudo apt-get install ca-certificates tree
      - name: install jellyfish
        run:  |
          wget https://github.com/gmarcais/Jellyfish/releases/download/v1.1.12/jellyfish-linux -O jellyfish1
          chmod -v +x jellyfish1
          ln -sv jellyfish1 jellyfish
          echo $(realpath .) >> $GITHUB_PATH
          tree $(realpath .)
      - name: install-edirect
        run:  |
          sh -c "$(curl -fsSL https://ftp.ncbi.nlm.nih.gov/entrez/entrezdirect/install-edirect.sh)"
          echo $HOME/edirect >> $GITHUB_PATH
          echo $GITHUB_WORKSPACE/Kalamari/bin >> $GITHUB_PATH
          tree $HOME/edirect
      - name: check-env
        run:  echo "$PATH"
      - name: select for only Listeria
        run:  |
          head -n 1 Kalamari/src/chromosomes.tsv > ${{ env.TSV }}
          grep -m 2 Listeria Kalamari/src/chromosomes.tsv >> ${{ env.TSV }}
          echo "These are the Listeria genomes for downstream tests"
          column -ts $'\t' ${{ env.TSV }}
          hexdump -c ${{ env.TSV }}
      - name: download
        run: |
          perl Kalamari/bin/downloadKalamari.pl --outdir ${{ env.OUTDIR }} ${{ env.TSV }}
          find ${{ env.OUTDIR }} -name '*.fasta.gz' | xargs gunzip -v
      - name: check-results
        run:  tree ${{ env.OUTDIR }}
      - name: install kraken
        run:  |
          wget https://github.com/DerrickWood/kraken/archive/refs/tags/v1.1.1.tar.gz -O kraken-v1.1.1.tar.gz
          tar zxvf kraken-v1.1.1.tar.gz
          cd kraken-1.1.1 && bash install_kraken.sh kraken-src && cd -
          chmod -v +x kraken-1.1.1/kraken-src/*
          echo $(realpath kraken-1.1.1/kraken-src) >> $GITHUB_PATH
          tree $(realpath) kraken-1.1.1
      - name: build taxonomy
        run: |
          export PATH=$PATH:Kalamari/bin
          buildTaxonomy.sh
          ls -lh Kalamari/share
      - name: Kraken1 database
        run:  |
          echo $PATH
          which kraken-build
          mkdir -pv kraken
          cp -rv Kalamari/share/kalamari-*/taxonomy kraken/taxonomy
          find ${{ env.OUTDIR }} -name '*.fasta' -exec kraken-build --db kraken --add-to-library {} \;
          tree kraken
          # Some super debugging here with -x
          #sed -i.bak 's/set -e/set -e; set -x/' $(which build_kraken_db.sh)
          #grep -v '^#' -m 20 -C 2 $a_bin/build_kraken_db.sh
          echo ".....Building the database....."
          kraken-build --minimizer-len 9 --rebuild --db kraken --threads 2 --work-on-disk
      - name: Kraken1 query
        run:  |
          QUERY=$(find ${{ env.OUTDIR }} -name '*.fasta' | head -n 1)
          echo "Query is $QUERY"
          kraken --db kraken -output kraken.raw --fasta-input $QUERY
      - name: Kraken1 view results
        run:  |
          tree kraken
          ls -lhSR kraken
          echo
          head kraken.raw