singularity-builds/assembly_improvement/assembly_improvement.def

Bootstrap: library
From: default/centos:latest

%post
  # Quiet perl down
  export LC_ALL=C

  # Install package files: repositories
  dnf install dnf-plugins-core -y
  dnf install epel-release -y
  dnf config-manager --set-enabled PowerTools

  # Operating dependencies
  dnf --setopt=install_weak_deps=False install procps perl python3 which -y

  # Building dependencies
  dnf groupinstall 'Development Tools' -y
  dnf install perl-App-cpanminus boost-devel cmake jsoncpp-devel jemalloc-devel expat-devel -y

  # Install SSPACE
  cd /opt || exit 1
  git clone https://github.com/nsoranzo/sspace_basic.git
  cd sspace_basic || exit 1
  git checkout 4fe5c # Not available as release tarball
  cd || exit 1
  export PATH=$PATH:/opt/sspace_basic

  # Install GapFiller
  curl -L https://downloads.sourceforge.net/project/gapfiller/v2.1.2/gapfiller-2.1.2.tar.gz | tar xvz
  cd gapfiller-2.1.2
  ./configure
  make && make install
  cd .. || exit 1
  rm -rf gapfiller-2.1.2

  # Install khmer
  pip3 install khmer

  # Install MUMmer
  curl -L https://github.com/mummer4/mummer/releases/download/v4.0.0rc1/mummer-4.0.0rc1.tar.gz | tar xvz
  cd mummer-4.0.0rc1 || exit 1
  ./configure
  make && make install
  cd .. || exit 1
  rm -rf mummer-4.0.0rc1

  # SGA uses an old C++ standard, so all of its dependencies must be compiled
  # against that standard, too
  export CFLAGS=-std=gnu++03
  export CXXFLAGS=-std=gnu++03
  export CPPFLAGS=-std=gnu++03

  # Install Google sparsehash (dependency for SGA)
  curl -L https://github.com/sparsehash/sparsehash/archive/refs/tags/sparsehash-2.0.4.tar.gz | tar xvz
  cd sparsehash-sparsehash-2.0.4 || exit 1
  ./configure
  make && make install
  cd .. || exit 1
  rm -rf sparsehash-sparsehash-2.0.4

  # Install bamtools library (dependency for SGA)
  curl -L https://github.com/pezmaster31/bamtools/archive/refs/tags/v2.5.2.tar.gz | tar xvz
  cd bamtools-2.5.2 || exit 1
  mkdir build && cd build || exit 1
  # Dependency issue unique to RHEL/CentOS 8: https://github.com/ComplianceAsCode/content/issues/7016#issuecomment-845066366
  dnf upgrade libarchive -y
  cmake ..
  make && make install
  cd ../.. || exit 1
  rm -rf bamtools-2.5.2

  # Install ruffus (dependency for SGA)
  pip3 install ruffus --upgrade

  # Install pysam (dependency for SGA)
  pip3 install pysam

  # Install SGA
  git clone https://github.com/jts/sga.git
  cd sga/src || exit 1
  git checkout 229293b # Contains compiler-blocking bug fixes that are not in the release tarball
  ./autogen.sh
  ./configure --with-bamtools=/usr/local
  make && make install
  cd ../.. || exit 1
  rm -rf sga

  # Unset the compiler flags
  export CFLAGS=''
  export CXXFLAGS=''
  export CPPFLAGS=''

  # QUASR is nowhere to be found, so skip installing it

  # Install assembly_improvement's dependencies
  cpanm -f Bio::AssemblyImprovement

  # Install an up-to-date version of assembly_improvement
  cd /opt || exit 1
  curl -L https://github.com/sanger-pathogens/assembly_improvement/releases/download/v2021.01.04.08.24.00.726/Bio_AssemblyImprovement-2021.01.04.08.24.00.726.tar.gz | tar xvz
  export PATH=$PATH:/opt/Bio_AssemblyImprovement-2021.01.04.08.24.00.726/bin
  cd || exit 1

  # Shim preloaded arguments into PATH
  IMPROVE_PREFIX=/opt/Bio_AssemblyImprovement-2021.01.04.08.24.00.726/bin
  SSPACE_PATH=/opt/sspace_basic/SSPACE_Basic.pl
  GF_PATH=/usr/local/bin/GapFiller
  ABACAS_PATH=/usr/local/bin/abacas.pl
  SGA_PATH=/usr/local/bin/sga
  cd /opt || exit 1
  mkdir shims && cd shims || exit 1
  echo "$IMPROVE_PREFIX/improve_assembly -s $SSPACE_PATH -g $GF_PATH -b $ABACAS_PATH \"\$@\"" > improve_assembly
  echo "$IMPROVE_PREFIX/diginorm_with_khmer -py /usr/bin/python3 \"\$@\"" > diginorm_with_khmer
  echo "$IMPROVE_PREFIX/fill_gaps_with_gapfiller -s $GF_PATH \"\$@\"" > fill_gaps_with_gapfiller
  echo "$IMPROVE_PREFIX/order_contigs_with_abacas -b $ABACAS_PATH \"\$@\"" > order_contigs_with_abacas
  echo "$IMPROVE_PREFIX/read_correction_with_sga -s $SGA_PATH \"\$@\"" > read_correction_with_sga
  echo "$IMPROVE_PREFIX/scaffold_with_sspace -s $SSPACE_PATH \"\$@\"" > scaffold_with_sspace
  echo "echo 'Error! QUASR is discontinued and not available in this container.'; exit 1" > remove_primers_with_quasr
  chmod +x ./*
  cd || exit 1
  export PATH=/opt/shims:$PATH

  # Clean up
  dnf groupremove 'Development Tools' -y
  dnf remove perl-App-cpanminus boost-devel cmake jsoncpp-devel jemalloc-devel expat-devel -y
  dnf autoremove -y
  dnf clean all

%environment
  export LC_ALL=C
  export PATH=$PATH:/opt/sspace_basic
  export PATH=$PATH:/opt/Bio_AssemblyImprovement-2021.01.04.08.24.00.726/bin
  export PATH=/opt/shims:$PATH

%test
  # We can use bash notation here, since /bin/sh is symlinked to bash in CentOS
  CMDS=('bash' 'ps' 'SSPACE_Basic.pl' 'GapFiller' 'abacas.pl' 'sga' 'improve_assembly' 'diginorm_with_khmer' 'fill_gaps_with_gapfiller' 'order_contigs_with_abacas' 'read_correction_with_sga' 'scaffold_with_sspace')
  for CMD in "${CMDS[@]}"; do
    if ! command -v "$CMD"; then
      echo "command $CMD not found!"
      FAILED_TESTS=1
    fi
  done
  # Abort if a test failed
  if [ -n "$FAILED_TESTS" ]; then
    exit 1
  fi

%runscript
  /opt/Bio_AssemblyImprovement-2021.01.04.08.24.00.726/bin/improve_assembly -s /opt/sspace_basic/SSPACE_Basic.pl -g /usr/local/bin/GapFiller -b /usr/local/bin/abacas.pl "$@"

%labels
  Author 25492070+MillironX@users.noreply.github.com
  SoftwareVersion 1.160490
  SingularityDefinitionVersion 2

%help
  Assembly Improvement
  ====================

  Take in an assembly in FASTA format, reads in FASTQ format, and make the
  assembly better by scaffolding and gap filling.

  **NOTE:** When using this container, **do not** use options that point to the
  paths of binaries. These have already been included in the container, and
  shims are setup to point to the containerized binaries already.
  E.g. `improve_assembly` is shimmed to execute
  `improve_assembly -s /opt/sspace_basic/SSPACE_Basic.pl -g /usr/local/bin/GapFiller -b /usr/local/bin/abacas.pl "$@"`,
  so providing `-s`, `-g`, or `-b` will break the usage.

  For more help, see
  - <https://github.com/sanger-pathogens/assembly_improvement>
  - <https://github.com/MillironX/singularity-builds>