Move minimize to simulation

devtools/conda-envs/macos-latest/spectrum.yaml

@@ -32,9 +32,6 @@ dependencies:
   - biopython
   - mdanalysis
   - openeye-toolkits
-  - openfe ~=1.0
-  - openmm
-  - gufe >=1.0.0
   - pymol-open-source
   - rdkit
   - plip

drugforge-alchemy/pyproject.toml

@@ -26,7 +26,7 @@ dependencies = []
 "Bug Tracker" = "https://github.com/choderalab/drugforge/issues"
 
 [project.scripts]
-asap-alchemy = "drugforge.alchemy.cli.cli:alchemy"
+drugforge-alchemy = "drugforge.alchemy.cli.cli:alchemy"
 
 [tool.setuptools.packages.find]
 where = ["."]

drugforge-cli/pyproject.toml

@@ -26,7 +26,7 @@ dependencies = []
 "Bug Tracker" = "https://github.com/choderalab/drugforge/issues"
 
 [project.scripts]
-asap-cli = "drugforge.cli.cli:cli"
+drugforge-cli = "drugforge.cli.cli:cli"
 
 [tool.setuptools.packages.find]
 where = ["."]

drugforge-data/drugforge/data/schema/ligand.py

@@ -47,7 +47,7 @@
 )
 
 if TYPE_CHECKING:
-    import openfe
+    import gufe
     from rdkit import Chem
 
 logger = logging.getLogger(__name__)
@@ -386,16 +386,16 @@ def to_rdkit(self) -> "Chem.Mol":
         set_SD_data(rdkit_mol, data)
         return rdkit_mol
 
-    def to_openfe(self) -> "openfe.SmallMoleculeComponent":
+    def to_openfe(self) -> "gufe.components.SmallMoleculeComponent":
         """
         Convert to an openfe SmallMoleculeComponent via the rdkit interface.
         """
-        import openfe
+        import gufe
 
-        return openfe.SmallMoleculeComponent.from_rdkit(self.to_rdkit())
+        return gufe.components.SmallMoleculeComponent.from_rdkit(self.to_rdkit())
 
     @classmethod
-    def from_openfe(cls, mol: "openfe.SmallMoleculeComponent", **kwargs) -> "Ligand":
+    def from_openfe(cls, mol: "gufe.components.SmallMoleculeComponent", **kwargs) -> "Ligand":
         """
         Create a Ligand from an openfe SmallMoleculeComponent
         """

drugforge-data/drugforge/data/testing/test_files.yaml

@@ -351,7 +351,7 @@ files:
   - resource: Mpro-YP_009725301_AFold_processed.pdb
     sha256hash: 3cdc93de0186243d790e7abd61d77eacde649f547f3d0cd0aaa58243631431f8
 
-  # asap-spectrum testing
+  # spectrum testing
   - resource: SARS_blast_results.xml
     sha256hash: d6e59d1f6b2eeee5162cca4dbeee7ffcd56eeeabdb733a505ff86ee1588cf8f6
 

drugforge-ml/pyproject.toml

@@ -26,7 +26,7 @@ dependencies = []
 "Bug Tracker" = "https://github.com/choderalab/drugforge/issues"
 
 [project.scripts]
-asap-ml = "drugforge.ml.cli:ml"
+drugforge-ml = "drugforge.ml.cli:ml"
 convert-ds = "drugforge.ml.scripts.convert_dataset:convert"
 
 [tool.setuptools.packages.find]

drugforge-simulation/drugforge/simulation/cli.py

@@ -13,12 +13,14 @@
     output_dir,
     pdb_file,
     use_dask,
+    target
 )
 from drugforge.data.readers.molfile import MolFileFactory
 from drugforge.data.schema.complex import Complex
 from drugforge.data.util.dask_utils import DaskType, make_dask_client_meta
 from drugforge.data.util.logging import FileLogger
 from drugforge.simulation.simulate import OpenMMPlatform, VanillaMDSimulator
+from drugforge.simulation.simulate import minimize_from_pdb
 from openmm import unit
 
 
@@ -144,3 +146,43 @@ def vanilla_md(
 @pdb_file
 def szybki():
     raise NotImplementedError("Szybki simulation not yet implemented")
+
+@simulation.command()
+@pdb_file
+@click.option(
+    "--min-out",
+    type=str,
+    default="minimized.pdb",
+    help="Optional file name for saving result of BLAST search",
+)
+@output_dir
+@md_openmm_platform
+@target
+@loglevel
+def minimize(
+    pdb_file: Union[Path, str],
+    min_out: Union[Path, str],
+    output_dir: Union[Path, str],
+    md_openmm_platform: str,
+    target: str,
+    loglevel: Union[int, str] = logging.INFO,
+):
+    logger = FileLogger(
+        "",  # default root logger so that dask logging is forwarded
+        path=output_dir,
+        logfile="vanilla_md.log",
+        stdout=True,
+        level=loglevel,
+    ).getLogger()
+
+    logger.info("Running minimization")
+
+    minimize_from_pdb(
+        pdb_file=pdb_file,
+        min_out=min_out,
+        output_dir=output_dir,
+        md_openmm_platform=md_openmm_platform,
+        target=target,
+    )
+    logger.info(f"Minimized structure saved to {min_out}")
+

drugforge-simulation/drugforge/simulation/simulate.py

@@ -1,5 +1,6 @@
 import abc
 import logging
+import os
 import shutil
 import warnings
 from pathlib import Path
@@ -653,3 +654,72 @@ def run_production_simulation(
 
     def provenance(self) -> dict[str, str]:
         return {}
+
+def minimize_from_pdb(
+    pdb_file: Union[Path, str],
+    min_out: Union[Path, str],
+    output_dir: Union[Path, str],
+    md_openmm_platform: str,
+    target: str,
+) -> Union[Path, str]:
+    """MD energy minimization a protein ligand complex. 
+    Energy minimization is performed with OpenMM (no equilibration or production MD is performed).
+
+    Parameters
+    ----------
+    pdb_complex : Union[Path, str]
+        Path to protein ligand complex pdb
+    min_out : Union[Path, str]
+        Output file with minimized pdb
+    out_dir : Union[Path, str]
+        Directory to save output minimized complex
+    md_platform : str
+        MD OpenMM platform [CPU, CUDA, OpenCL, Reference, Fastest]
+    target : str
+        Name of reference target (see drugforge documentation).
+
+    Returns
+    -------
+    str
+       Path to minimized file
+    """
+    from drugforge.data.schema.complex import Complex
+    from drugforge.modeling.schema import PreppedComplex
+
+    if Path(min_out).is_file():
+        logger.warning(
+            f"The file {min_out} already exists. The minimization will be skipped"
+        )
+        return min_out
+    comp_name = "MOL"
+    cmp = Complex.from_pdb(
+        pdb_file,
+        ligand_kwargs={"compound_name": comp_name},
+        target_kwargs={"target_name": target},
+    )
+    out_dir = Path(output_dir)
+    prepped_cmp = PreppedComplex.from_complex(cmp)
+    prepped_cmp.target.to_pdb_file(out_dir / "target.pdb")
+    cmp.ligand.to_sdf(out_dir / "ligand.sdf")
+
+    md_simulator = VanillaMDSimulator(
+        output_dir=out_dir,
+        openmm_platform=md_openmm_platform,
+        minimize_only=True,
+        num_steps=1,
+    )
+    simulation_results = md_simulator.simulate(
+        [(out_dir / "target.pdb", out_dir / "ligand.sdf")],
+        outpaths=[out_dir],
+        failure_mode="skip",
+    )
+    min_path = simulation_results[0].minimized_pdb_path
+    shutil.move(min_path, min_out)
+    shutil.rmtree(f"{out_dir}/target_ligand", ignore_errors=True)
+    for file_path in [f"{out_dir}/target.pdb", f"{out_dir}/ligand.sdf"]:
+        try:
+            os.remove(file_path)
+        except FileNotFoundError:
+            pass
+
+    return min_out

drugforge-simulation/drugforge/simulation/tests/conftest.py

@@ -12,7 +12,7 @@ def ligand_path():
 
 @pytest.fixture(scope="session")
 def protein_path():
-    return fetch_test_file("Mpro-x0071_0A_ERI-UCB-8c6b7d0d-1.pdb")
+    return fetch_test_file("Mpro-P2660_0A_bound.pdb")
 
 
 @pytest.fixture(scope="session")

drugforge-simulation/drugforge/simulation/tests/test_simulate.py

@@ -3,6 +3,7 @@
 from unittest import mock
 
 import pytest
+from pathlib import Path
 from drugforge.docking.docking import DockingResult
 from drugforge.docking.openeye import POSITDockingResults
 from drugforge.simulation.cli import simulation as cli
@@ -201,3 +202,27 @@ def _simulate_patch(
         ]
         result = runner.invoke(cli, args)
     assert click_success(result)
+
+
+
+@pytest.mark.skipif(os.getenv("SKIP_EXPENSIVE_TESTS"), reason="Expensive tests skipped")
+def test_minimize_cli(protein_path, tmp_path):
+    """Test minimization of protein PDB using OpenMM."""
+    runner = CliRunner()
+    min_out = f"{tmp_path}/min_out.pdb"
+    args = [
+        "minimize",
+        "--pdb-file",
+        protein_path,
+        "--min-out",
+        min_out,
+        "--output-dir",
+        tmp_path,
+        "--md-openmm-platform",
+        "CPU",
+        "--target",
+        "SARS-CoV-2-Mpro",
+    ]
+    result = runner.invoke(cli, args)
+    assert click_success(result)
+    assert Path(min_out).exists()

drugforge-spectrum/drugforge/spectrum/score.py

@@ -12,7 +12,6 @@
 )
 from drugforge.data.backend.openeye import oechem, split_openeye_mol
 from drugforge.spectrum.calculate_rmsd import rmsd_alignment
-from drugforge.simulation.simulate import VanillaMDSimulator
 from drugforge.data.services.postera.manifold_data_validation import TargetTags
 from drugforge.data.metadata.resources import active_site_chains
 
@@ -641,73 +640,3 @@ def score_gnina(pdb_target:str,
             ]
         )
     return df
-
-
-def minimize_structure(
-    pdb_complex: Union[Path, str],
-    min_out: Union[Path, str],
-    out_dir: Union[Path, str],
-    md_platform: str,
-    comp_name: str,
-    target_name: str,
-) -> Union[Path, str]:
-    """MD energy minimization a protein ligand complex. 
-    Energy minimization is performed with OpenMM (no equilibration or production MD is performed).
-
-    Parameters
-    ----------
-    pdb_complex : Union[Path, str]
-        Path to protein ligand complex pdb
-    min_out : Union[Path, str]
-        Output file with minimized pdb
-    out_dir : Union[Path, str]
-        Directory to save output minimized complex
-    md_platform : str
-        MD OpenMM platform [CPU, CUDA, OpenCL, Reference, Fastest]
-    comp_name : str
-        Name of to give to complex. Can be arbitrary.
-    target_name : str
-        Name of reference target (see drugforge documentation).
-
-    Returns
-    -------
-    str
-       Path to minimized file
-    """
-
-    if Path(min_out).is_file():
-        logger.warning(
-            f"The file {min_out} already exists. The minimization will be skipped"
-        )
-        return min_out
-    cmp = Complex.from_pdb(
-        pdb_complex,
-        ligand_kwargs={"compound_name": comp_name},
-        target_kwargs={"target_name": target_name},
-    )
-    out_dir = Path(out_dir)
-    prepped_cmp = PreppedComplex.from_complex(cmp)
-    prepped_cmp.target.to_pdb_file(out_dir / "target.pdb")
-    cmp.ligand.to_sdf(out_dir / "ligand.sdf")
-
-    md_simulator = VanillaMDSimulator(
-        output_dir=out_dir,
-        openmm_platform=md_platform,
-        minimize_only=True,
-        num_steps=1,
-    )
-    simulation_results = md_simulator.simulate(
-        [(out_dir / "target.pdb", out_dir / "ligand.sdf")],
-        outpaths=[out_dir],
-        failure_mode="skip",
-    )
-    min_path = simulation_results[0].minimized_pdb_path
-    shutil.move(min_path, min_out)
-    shutil.rmtree(f"{out_dir}/target_ligand", ignore_errors=True)
-    for file_path in [f"{out_dir}/target.pdb", f"{out_dir}/ligand.sdf"]:
-        try:
-            os.remove(file_path)
-        except FileNotFoundError:
-            pass
-
-    return min_out

drugforge-spectrum/drugforge/spectrum/tests/test_score.py

@@ -7,7 +7,6 @@
     dock_and_score,
     get_ligand_rmsd,
     score_autodock_vina,
-    minimize_structure
 )
 from drugforge.spectrum.calculate_rmsd import get_binding_site_rmsd
 from drugforge.docking.scorer import ChemGauss4Scorer
@@ -105,18 +104,3 @@ def test_vina_score(target_prepped_vina, ligand_prepped_vina):
         dock=False,
     )
     assert df_vina["Vina-score-premin"].values[0] < 0
-
-
-@pytest.mark.skipif(os.getenv("SKIP_EXPENSIVE_TESTS"), reason="Expensive tests skipped")
-def test_minimize(protein_path, tmp_path):
-    """Test minimization of protein PDB using OpenMM."""
-    min_out = f"{tmp_path}/min_out.pdb"
-    minimize_structure(
-    pdb_complex = protein_path,
-    min_out = min_out,
-    out_dir = tmp_path,
-    md_platform = 'CPU',
-    comp_name = 'Mol',
-    target_name = 'SARS-CoV-2',)
-
-    assert Path(min_out).exists()

drugforge-workflows/drugforge/workflows/spectrum_workflows/cli.py

@@ -564,7 +564,7 @@ def fitness_alignment(
     "--ml-score",
     is_flag=True,
     default=False,
-    help="Whether to employ asap-implemented ML models to score poses.",
+    help="Whether to employ implemented ML models to score poses.",
 )
 @click.option(
     "--bsite-rmsd",