refactor pose_cluster_filter

utils/docking/diffdock/.gitattributes

@@ -0,0 +1,5 @@
+*.pdb filter=lfs diff=lfs merge=lfs -text
+*.pdbqt filter=lfs diff=lfs merge=lfs -text
+*.mol2 filter=lfs diff=lfs merge=lfs -text
+*.xlsx filter=lfs diff=lfs merge=lfs -text
+*.sdf filter=lfs diff=lfs merge=lfs -text

utils/pose-cluster-filter-plugin/Dockerfile → utils/docking/diffdock/pose-cluster-filter-tool/Dockerfile

@@ -1,5 +1,5 @@
 
-# docker build -f Dockerfile -t mrbrandonwalker/pose_cluster_filter_tool .
+# docker build -f Dockerfile -t polusai/pose-cluster-filter-tool .
 FROM condaforge/mambaforge
 
 ENV EXEC_DIR="/opt/executables"
@@ -16,16 +16,11 @@ COPY README.md ${EXEC_DIR}
 COPY CHANGELOG.md ${EXEC_DIR}
 
 # Install needed packages here
-# errors installing rdkit from poetry so using conda
-COPY environment.yml ${EXEC_DIR}
-RUN mamba env create -f ${EXEC_DIR}/environment.yml
-RUN echo "source activate project_env" > ~/.bashrc
-ENV PATH /opt/conda/envs/env/bin:$PATH
 
 COPY src ${EXEC_DIR}/src
 
 ADD Dockerfile .
 
-RUN conda run -n project_env pip install ${EXEC_DIR} --no-cache-dir
+RUN pip3 install ${EXEC_DIR} --no-cache-dir
 
 CMD ["--help"]

utils/docking/diffdock/pose-cluster-filter-tool/build-docker.sh

@@ -0,0 +1,4 @@
+#!/bin/bash
+
+version=$(<VERSION)
+docker build . -t polusai/pose-cluster-filter-tool:${version}

utils/pose-cluster-filter-plugin/ict.yml → utils/docking/diffdock/pose-cluster-filter-tool/ict.yml

@@ -1,12 +1,12 @@
 specVersion: "0.1.0"
 name: pose_cluster_filter
 version: 0.1.0
-container: pose-cluster-filter-plugin
+container: pose-cluster-filter-tool
 entrypoint:
 title: pose_cluster_filter
 description: Cluster poses in protein and take max confident pose for each cluster
-author: Data Scientist
-contact: [email protected]
+author: Brandon Walker, Nazanin Donyapour
+contact: [email protected], [email protected]
 repository:
 documentation:
 citation:

utils/pose-cluster-filter-plugin/pose_cluster_filter.cwl → utils/docking/diffdock/pose-cluster-filter-tool/pose_cluster_filter_0@1@0.cwl

@@ -8,11 +8,11 @@ label: Cluster poses in protein and take max confident pose for each cluster
 doc: |-
   Cluster poses in protein and take max confident pose for each cluster
 
-baseCommand: ["conda", "run", "-n", "project_env", "python", "-m", "polus.mm.utils.pose_cluster_filter"]
+baseCommand: ["python", "-m", "polus.mm.utils.pose_cluster_filter"]
 
 hints:
   DockerRequirement:
-    dockerPull: mrbrandonwalker/pose_cluster_filter_tool
+    dockerPull: polusai/pose-cluster-filter-tool@sha256:f30068018bf5e04f6c19ea9fe44349066732720d0307d0a569dabb98d199312e
 
 requirements:
   InlineJavascriptRequirement: {}
@@ -30,8 +30,8 @@ inputs:
     type:
       type: array
       items: File
-      inputBinding:
-        prefix: --predicted_poses
+    inputBinding:
+      prefix: --predicted_poses
     format: edam:format_3814
 
 outputs:

utils/pose-cluster-filter-plugin/pyproject.toml → utils/docking/diffdock/pose-cluster-filter-tool/pyproject.toml

@@ -7,7 +7,10 @@ readme = "README.md"
 packages = [{include = "polus", from = "src"}]
 
 [tool.poetry.dependencies]
+python = ">=3.9,<3.12"
 typer = "^0.7.0"
+sophios = "0.1.1"
+rdkit = "2024.3.3"
 
 [tool.poetry.group.dev.dependencies]
 bump2version = "^1.0.1"

utils/pose-cluster-filter-plugin/src/polus/mm/utils/pose_cluster_filter/__main__.py → utils/docking/diffdock/pose-cluster-filter-tool/src/polus/mm/utils/pose_cluster_filter/__main__.py

@@ -1,13 +1,14 @@
 """Package entrypoint for the pose_cluster_filter package."""
 
 # Base packages
+import argparse
 import logging
 from os import environ
 from pathlib import Path
 
-import typer
 from polus.mm.utils.pose_cluster_filter.pose_cluster_filter import pose_cluster_filter
 
+# Set up logging
 logging.basicConfig(
     format="%(asctime)s - %(name)-8s - %(levelname)-8s - %(message)s",
     datefmt="%d-%b-%y %H:%M:%S",
@@ -16,31 +17,44 @@
 logger = logging.getLogger("polus.mm.utils.pose_cluster_filter.")
 logger.setLevel(POLUS_LOG)
 
-app = typer.Typer(help="pose_cluster_filter.")
 
+def main() -> None:
+    """Parse the arguments and call the function."""
+    # Create the argument parser
+    parser = argparse.ArgumentParser(description="pose_cluster_filter.")
 
-@app.command()
-def main(
-    centroid_cutoff: float = typer.Option(
-        ...,
+    # Add arguments
+    parser.add_argument(
         "--centroid_cutoff",
-        help="",
-    ),
-    predicted_poses: list[Path] = typer.Option(
-        ...,
+        type=float,
+        required=True,
+        help="The cutoff distance for the centroid.",
+    )
+    parser.add_argument(
         "--predicted_poses",
-        help="",
-    ),
-) -> None:
-    """pose_cluster_filter."""
+        type=Path,
+        nargs="+",
+        required=True,
+        help="A list of paths to the predicted poses.",
+    )
+
+    # Parse the arguments
+    args = parser.parse_args()
+
+    # Extract arguments
+    centroid_cutoff = args.centroid_cutoff
+    predicted_poses = args.predicted_poses
+
+    # Log the arguments
     logger.info(f"centroid_cutoff: {centroid_cutoff}")
     logger.info(f"predicted_poses: {predicted_poses}")
 
+    # Call the function with the arguments
     pose_cluster_filter(
         centroid_cutoff=centroid_cutoff,
         predicted_poses=predicted_poses,
     )
 
 
 if __name__ == "__main__":
-    app()
+    main()

utils/pose-cluster-filter-plugin/tests/test_pose_cluster_filter.py → utils/docking/diffdock/pose-cluster-filter-tool/tests/test_pose_cluster_filter.py

@@ -1,16 +1,9 @@
 """Tests for pose_cluster_filter."""
-import sys
 from pathlib import Path
 
 from polus.mm.utils.pose_cluster_filter.pose_cluster_filter import pose_cluster_filter
-
-current_dir = Path(__file__).resolve().parent
-target_dir = current_dir.parent.parent.parent / "cwl_utils"
-sys.path.append(str(target_dir))
-
-from cwl_utilities import call_cwltool  # noqa: E402
-from cwl_utilities import create_input_yaml  # noqa: E402
-from cwl_utilities import parse_cwl_arguments  # noqa: E402
+from sophios.api.pythonapi import Step
+from sophios.api.pythonapi import Workflow
 
 
 def test_pose_cluster_filter() -> None:
@@ -35,22 +28,38 @@ def test_pose_cluster_filter() -> None:
 
 def test_pose_cluster_filter_cwl() -> None:
     """Test the pose_cluster_filter CWL."""
-    cwl_file = Path("pose_cluster_filter.cwl")
-    input_to_props = parse_cwl_arguments(cwl_file)
+    cwl_file_str = "pose_cluster_filter_0@[email protected]"
+    cwl_file = Path(__file__).resolve().parent.parent / Path(cwl_file_str)
+
     predicted_poses = [
         "rank2_confidence0.35.sdf",
         "rank3_confidence0.34.sdf",
         "rank1_confidence0.36.sdf",
     ]
-    file_dict = input_to_props["predicted_poses"][0]
-    input_to_props["predicted_poses"] = []
-    for pose in predicted_poses:
-        file_dict_current = file_dict.copy()
-        path_pose = str(Path(__file__).resolve().parent / Path(pose))
-        file_dict_current["path"] = path_pose
-        input_to_props["predicted_poses"].append(file_dict_current)
-
-    input_yaml_path = Path("pose_cluster_filter.yml")
-    create_input_yaml(input_to_props, input_yaml_path)
-    stdout, stderr = call_cwltool(cwl_file, input_yaml_path)
-    assert "success" in stderr
+
+    pose_cluster_filter = Step(clt_path=cwl_file)
+    pose_cluster_filter.predicted_poses = [
+        {
+            "format": "edam:format_3814",
+            "class": "File",
+            "path": str(Path(__file__).resolve().parent / Path(pose)),
+        }
+        for pose in predicted_poses
+    ]
+    pose_cluster_filter.centroid_cutoff = 5
+
+    steps = [pose_cluster_filter]
+    filename = "pose_cluster_filter"
+    viz = Workflow(steps, filename)
+
+    viz.run()
+
+    outdir = Path("outdir")
+
+    # Use rglob to find all .sdf files
+    sdf_files = list(outdir.rglob("*.sdf"))
+
+    # Check if any .sdf files were found
+    if not sdf_files:
+        msg = f"No .sdf files found in '{outdir}'."
+        raise FileNotFoundError(msg)