add bbmap tool

.github/workflows/build-push-bbmap-image.yml

@@ -0,0 +1,31 @@
+name: Build & Push BBMap Image to GHCR
+
+on:
+  pull_request:
+    types:
+      - opened
+      - reopened
+      - synchronize
+      - ready_for_review
+    paths:
+      - 'src/loaders/compute_tools/bbmap/versions.yaml'
+      - '.github/workflows/build-push-bbmap-image.yml'
+      - '.github/workflows/build-push-tool-images.yml'
+
+  push:
+    branches:
+      - main
+      - master
+      - develop
+    paths:
+      - 'src/loaders/compute_tools/bbmap/versions.yaml'
+      - '.github/workflows/build-push-bbmap-image.yml'
+      - '.github/workflows/build-push-tool-images.yml'
+
+jobs:
+  trigger-build-push:
+    uses: ./.github/workflows/build-push-tool-images.yml
+    with:
+      tool_name: bbmap
+      version_file: 'src/loaders/compute_tools/bbmap/versions.yaml'
+    secrets: inherit

src/loaders/compute_tools/bbmap/Dockerfile

@@ -0,0 +1,31 @@
+FROM continuumio/miniconda3:24.5.0-0
+
+# NOTE: If the tool version changes, ensure the metadata information saved after running the tool in the _run_bbmap_single method is updated
+ARG BBMAP_VER=39.06
+ENV CONDA_ENV bbmap-$BBMAP_VER
+
+# Add Bioconda and Conda-Forge channels
+RUN conda config --add channels bioconda
+RUN conda config --add channels conda-forge
+
+# Install BBMap
+ARG PYTHON_VER=3.11
+RUN conda create -n $CONDA_ENV python=$PYTHON_VER bbmap=$BBMAP_VER
+
+# Activate the environment
+RUN echo "source activate $CONDA_ENV" >> ~/.bashrc
+
+# Set up directories
+RUN mkdir -p /app
+COPY ./ /app/collections
+RUN rm -r /app/collections/.git
+
+
+ENV PYTHONPATH /app/collections
+WORKDIR /app
+
+ENV PY_SCRIPT=/app/collections/src/loaders/compute_tools/bbmap/bbmap.py
+
+RUN chmod -R 777 /app/collections
+
+ENTRYPOINT ["/app/collections/src/loaders/compute_tools/entrypoint.sh"]

src/loaders/compute_tools/bbmap/bbmap.py

@@ -0,0 +1,50 @@
+"""
+Run BBMap tool on a set of faa files.
+
+This tool serves a distinct purpose separate from collection tools; instead, it is suited for CDM work.
+Therefore, the parser program is not compatible with data generated by this tool.
+
+"""
+from pathlib import Path
+
+from src.loaders.common.loader_common_names import TOOL_METADATA
+from src.loaders.compute_tools.tool_common import ToolRunner, run_command, create_tool_metadata
+
+
+def _run_bbmap_single(
+        tool_safe_data_id: str,
+        data_id: str,
+        source_file: Path,
+        output_dir: Path,
+        threads_per_tool_run: int,
+        debug: bool) -> None:
+
+    metadata_file = output_dir / TOOL_METADATA
+    if metadata_file.exists():
+        print(f"Skipping {source_file} as it has already been processed.")
+        return
+
+    command = [
+        'stats.sh',
+        'in=' + str(source_file),
+        'out=' + str(output_dir / 'result.json'),
+    ]
+
+    run_command(command, output_dir if debug else None)
+
+    # Save run info to a metadata file in the output directory for parsing later
+    metadata = {'tool_name': 'bbmap',
+                'version': '39.06',
+                'command': command,
+                'data_id': data_id,
+                'source_file': str(source_file),}
+    create_tool_metadata(output_dir, metadata)
+
+
+def main():
+    runner = ToolRunner("bbmap")
+    runner.parallel_single_execution(_run_bbmap_single, unzip=True)
+
+
+if __name__ == "__main__":
+    main()

src/loaders/compute_tools/bbmap/versions.yaml

@@ -0,0 +1,5 @@
+versions:
+  - version: 0.1.0
+    date: 2024-08-16
+    notes: |
+      - initial BBMap implementation

src/loaders/jobs/taskfarmer/task_generator.py

@@ -41,7 +41,7 @@
 
 '''
 
-TOOLS_AVAILABLE = ['gtdb_tk', 'checkm2', 'microtrait', 'mash', 'eggnog']
+TOOLS_AVAILABLE = ['gtdb_tk', 'checkm2', 'microtrait', 'mash', 'eggnog', 'bbmap']
 
 NODE_TIME_LIMIT_DEFAULT = 5  # hours
 # Used as THREADS variable in the batch script which controls the number of parallel tasks per node