Singularity.def

Bootstrap: docker
From: ubuntu:22.04

# ============================================================================
# RaPDTool v2.3.2 - Rapid Profiling and Deconvolution Tool for metagenomes
# ----------------------------------------------------------------------------
# Reproducible build recipe. Build with:
#     apptainer build --fakeroot rapdtool.sif Singularity.def
#
# NOTE: neither database is bundled, to keep the image small. Provide them at
# run time (see README):
#   - mash reference  ->  -d $RTMASHDB
#   - FOCUS k-mer db  ->  --focus-db $RTFOCUSDB   (a dir containing db/k6)
# ============================================================================

%files
    # pipeline code (paths relative to this .def)
    bin/rapdtool.py            /usr/local/bin/rapdtool.py
    bin/rapdtool_split_bins.py /usr/local/bin/rapdtool_split_bins.py
    bin/rapdtool_results.pl    /usr/local/bin/rapdtool_results.pl

%post
    export DEBIAN_FRONTEND=noninteractive
    apt-get update && apt-get install -y --no-install-recommends \
        wget ca-certificates git make gcc bzip2 procps
    rm -rf /var/lib/apt/lists/*

    # ---- Miniforge in /opt/miniconda3 -------------------------------------
    # Miniforge (community conda-forge installer): its base env has no Anaconda
    # 'defaults' channel and no conda-anaconda-tos package (avoids the defaults ToS
    # gate and the base-env python 3.14 conflict of the latest Miniconda).
    wget -q https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-Linux-x86_64.sh -O /tmp/mc.sh
    bash /tmp/mc.sh -b -p /opt/miniconda3 && rm /tmp/mc.sh
    export PATH=/opt/miniconda3/bin:$PATH

    # ---- Tools in a DEDICATED env (isolated from base conda/mamba lib pins) ----
    # The 2024-era tool builds fail to install into base because Miniforge's base
    # mamba pins modern libcurl/xz/zlib that conflict with metabat2=2.15. A fresh
    # env lets the solver pick the known-good combination.
    # NOTE: setuptools<81 keeps pkg_resources, which focus_app imports (removed in 81+).
    conda create -y -n rt --override-channels -c conda-forge -c bioconda \
        python=3.12 'setuptools<81' focus=1.8 metabat2=2.15 mash=2.3 hmmer=3.4 \
        prodigal=2.6.3 entrez-direct=21.6 perl-app-cpanminus
    ENV=/opt/miniconda3/envs/rt
    export PATH=$ENV/bin:$PATH

    # python tools into the env
    $ENV/bin/pip install --no-cache-dir miComplete==1.1.1 Binning_refiner==1.4.3

    # Patch miComplete for modern Biopython: Bio.SeqUtils.GC was removed in Biopython
    # 1.80+ (this env has newer), replaced by gc_fraction. Without this, miComplete
    # crashes on import and full mode leaves Completeness/Redundancy empty.
    PSEQ=$(find $ENV/lib -path '*/micomplete/parseseqs.py' | head -1)
    sed -i 's/from Bio.SeqUtils import GC/from Bio.SeqUtils import gc_fraction/; s/round(GC(/round(gc_fraction(/' "$PSEQ"

    # FOCUS ships a ~216 MB k-mer database (db/k6) inside its package; remove it
    # so it can be provided externally via --focus-db/$RTFOCUSDB.
    find $ENV -type d -path '*/focus_app/db' -exec rm -rf {} + 2>/dev/null || true

    # ---- KronaTools (text import; no taxonomy DB needed) -------------------
    git clone https://github.com/marbl/Krona /opt/Krona || true
    if [ -d /opt/Krona/KronaTools ]; then
        cd /opt/Krona/KronaTools && ./install.pl --prefix /opt/KronaTools || true
    fi

    # ---- Perl reporting module into the env's perl -----------------------
    # Not on conda; installed from CPAN via cpanm (needs 'make', installed above).
    # Version pinned for reproducibility. NO '|| true' here — a failure must abort.
    # rapdtool_results.pl sizes its own columns since v2.3.2, so plain
    # Text::SimpleTable suffices (::AutoWidth also pulled Moo and Type::Tiny).
    $ENV/bin/cpanm --notest Text::SimpleTable@2.07

    chmod +x /usr/local/bin/rapdtool.py /usr/local/bin/rapdtool_split_bins.py \
             /usr/local/bin/rapdtool_results.pl

    # ---- Self-check: FAIL the build if any essential component is missing --
    # (the install steps above tolerate transient issues; this gate ensures we never
    #  ship a "successful" but broken image — missing Krona / Perl module / focus_app)
    export PATH=/opt/KronaTools/bin:$PATH   # KronaTools installs here (mirrors runtime %environment)
    command -v ktImportText >/dev/null || { echo 'BUILD ERROR: KronaTools (ktImportText) missing'; exit 1; }
    $ENV/bin/perl -MText::SimpleTable -e1 || { echo 'BUILD ERROR: Perl module Text::SimpleTable missing'; exit 1; }
    $ENV/bin/python -c 'import focus_app.focus' || { echo 'BUILD ERROR: focus_app import failed (pkg_resources?)'; exit 1; }
    for t in focus metabat2 mash miComplete Binning_refiner esearch; do
        command -v "$t" >/dev/null || { echo "BUILD ERROR: $t missing"; exit 1; }
    done
    # miComplete/Binning_refiner must actually import & run (not just be on PATH)
    miComplete --help >/dev/null 2>&1 || { echo 'BUILD ERROR: miComplete not runnable (Biopython GC/gc_fraction?)'; exit 1; }
    Binning_refiner --help >/dev/null 2>&1 || { echo 'BUILD ERROR: Binning_refiner not runnable'; exit 1; }

    # ---- Slim the image (safe, runtime-only footprint) --------------------
    conda clean -a -y || true
    MC=/opt/miniconda3
    SP=$ENV/lib/python3.12/site-packages
    rm -rf $MC/pkgs /tmp/mc.sh
    find $ENV -name '*.a' -delete 2>/dev/null || true
    find $ENV -name '*.o' -delete 2>/dev/null || true
    find $MC -name '__pycache__' -type d -prune -exec rm -rf {} + 2>/dev/null || true
    # test suites, type stubs and sample data of heavy python pkgs (scipy/numpy/matplotlib/Bio...)
    find $SP -type d -name tests -prune -exec rm -rf {} + 2>/dev/null || true
    find $SP/numpy $SP/scipy -name '*.pyi' -delete 2>/dev/null || true
    rm -rf $SP/matplotlib/mpl-data/sample_data $SP/pip $SP/pip-*.dist-info
    # HMMER easel source tree, conda docs/man/info/locale
    rm -rf $ENV/share/easel $ENV/share/doc $ENV/share/man $MC/share/doc $MC/share/man $MC/share/info
    # unused stdlib + tkinter/tcl-tk (matplotlib runs headless via MPLBACKEND=Agg)
    rm -rf $ENV/lib/python3.12/idlelib $ENV/lib/python3.12/ensurepip \
           $ENV/lib/python3.12/lib2to3 $ENV/lib/python3.12/test \
           $ENV/lib/python3.12/tkinter $ENV/lib/python3.12/lib-dynload/_tkinter*.so \
           $ENV/lib/tcl8 $ENV/lib/tcl8.6 $ENV/lib/tk8.6 $ENV/lib/libtcl8.6.so $ENV/lib/libtk8.6.so
    apt-get clean
    rm -rf /var/lib/apt/lists/* /var/cache/apt/* /usr/share/doc/* /var/log/*

%environment
    export PATH=/opt/miniconda3/envs/rt/bin:/opt/KronaTools/bin:/usr/local/bin:$PATH
    export LC_ALL=C
    export MPLBACKEND=Agg          # headless matplotlib (no Tk/X); lets tkinter/tcl-tk be dropped

%runscript
    if [ "$#" -eq 0 ]; then
        exec rapdtool.py --help
    fi
    exec rapdtool.py "$@"

%labels
    Author Karel Estrada and Ayixon Sanchez-Reyes; pipeline maintained by the RaPDTool team
    Version 2.3.2
    Description RaPDTool metagenome profiling/binning/classification pipeline

%help
    RaPDTool v2.3.2
    Usage inside the container:
        rapdtool.py -i assembly.fasta --focus-db $RTFOCUSDB [-d $RTMASHDB]
                    [-o out_dir] [-m full|profile|screen] [-t threads]
                    [-a coverage.txt] [--no-split-bins]
        (full/screen need -d/$RTMASHDB; profile uses it only if given)

    The databases are NOT included. Provide them via environment variables:
        export RTMASHDB=/path/to/mash_database.msh
        export RTFOCUSDB=/path/to/focus     # a directory containing db/k6

# ----------------------------------------------------------------------------
# A prebuilt image is also attached to the GitHub Releases page; see the README.
# ----------------------------------------------------------------------------