Revert "Revert "Merge pull request PolusAI#230 from ndonyapour/Molgan_DockingRescoring_errorlogging""

This reverts commit a4e678c.

Author: Brandon Duane Walker

cwl_adapters/convert_mol2.cwl

@@ -126,6 +126,13 @@ outputs:
       glob: $(inputs.output_mol2_path)
     format: edam:format_3816 # mol2
 
+  stderr:
+    type: File
+    outputBinding:
+      glob: stderr
+
+stderr: stderr
+
 $namespaces:
   edam: https://edamontology.org/
 

cwl_adapters/extract_molecules_sdf.cwl

@@ -70,7 +70,7 @@ arguments: [$(inputs.input_path), "-o", "sdf", "-O", $(inputs.output_sdf_path),
 
 hints:
   DockerRequirement:
-    #dockerPull: jakefennick/openbabel
+    #dockerPull: ndonyapour/openbabel
     dockerPull: quay.io/biocontainers/biobb_chemistry:4.0.0--pyhdfd78af_1
 
 inputs:

cwl_adapters/load_trained_molgan_model.cwl

@@ -11,6 +11,12 @@ hints:
   DockerRequirement:
     dockerPull: ndonyapour/molgan
 
+# Set environment variables for the tool,
+# See: https://www.commonwl.org/user_guide/topics/environment-variables.html
+requirements:
+  EnvVarRequirement:
+    envDef:
+      RDKIT_ERROR_LOGGING: $(inputs.rdkit_error_logging)
 inputs:
   input_data_path:
     label: Path to the input data file
@@ -41,7 +47,7 @@ inputs:
     default: NP.gz
     inputBinding:
       prefix: --input_NP_Score_path
- 
+
   input_SA_Score_path:
     label: Output ceout file (AMBER ceout)
     doc: |-
@@ -60,10 +66,10 @@ inputs:
   input_model_dir:
     label: Input directory of trained models
     type: string
-    format: 
+    format:
     - edam:format_2330 # 'Textual format'
     inputBinding:
-      prefix: --input_model_dir 
+      prefix: --input_model_dir
     default: output
 
   output_log_path:
@@ -107,6 +113,14 @@ inputs:
       prefix: --num_samples
     default: 1000
 
+  rdkit_error_logging:
+    label: Enable or disable RDKit error logging
+    doc: |-
+      Enable or disable RDKit error logging
+    type: string?
+    # RDKit prints out all errors by default, which can pose issues for CI,
+    # particularly with large databases. It would be more efficient to suppress these errors.
+    default: "ON"
 outputs:
   output_log_path:
     label: Path to the log file
@@ -126,6 +140,12 @@ outputs:
       glob: $(inputs.output_sdf_path)
     format: edam:format_3814 # sdf
 
+  stderr:
+    type: File
+    outputBinding:
+      glob: stderr
+
+stderr: stderr
 
 $namespaces:
   edam: https://edamontology.org/

cwl_adapters/obmin.cwl

@@ -12,7 +12,7 @@ baseCommand: bash
 
 hints:
   DockerRequirement:
-    dockerPull: jakefennick/openbabel
+    dockerPull: ndonyapour/openbabel
 
 inputs:
   script:
@@ -31,8 +31,8 @@ inputs:
     type: string
     format:
     - edam:format_3816
-#    inputBinding:
-#      position: 3
+    inputBinding:
+       position: 3
     default: system.mol2
 
 outputs:
@@ -43,7 +43,13 @@ outputs:
     outputBinding:
       glob: $(inputs.output_mol2_path)
 
-stdout: $(inputs.output_mol2_path)
+  stderr:
+    type: File
+    outputBinding:
+      glob: stderr
+
+stderr: stderr
+
 
 $namespaces:
   edam: https://edamontology.org/

cwl_adapters/preprocess_molgan_database.cwl

@@ -11,6 +11,13 @@ hints:
   DockerRequirement:
     dockerPull: ndonyapour/molgan
 
+# Set environment variables for the tool,
+# See: https://www.commonwl.org/user_guide/topics/environment-variables.html
+requirements:
+  EnvVarRequirement:
+    envDef:
+      RDKIT_ERROR_LOGGING: $(inputs.rdkit_error_logging)
+
 inputs:
   input_sdf_path:
     label: Path to the input file
@@ -41,6 +48,17 @@ inputs:
       prefix: --output_data_path
     default: system.pkl
 
+  rdkit_error_logging:
+    label: Enable or disable RDKit error logging
+    doc: |-
+      Enable or disable RDKit error logging
+    type: string?
+    format:
+    - edam:format_2330
+    # RDKit prints out all errors by default, which can pose issues for CI,
+    # particularly with large databases. It would be more efficient to suppress these errors.
+    default: "ON"
+
 outputs:
   output_data_path:
     label: Path to the output data file
@@ -51,6 +69,13 @@ outputs:
       glob: $(inputs.output_data_path)
     format: edam:format_3653 # sdf
 
+  stderr:
+    type: File
+    outputBinding:
+      glob: stderr
+
+stderr: stderr
+
 $namespaces:
   edam: https://edamontology.org/
 

cwl_adapters/run_trained_molgan_model.cwl

@@ -3,14 +3,21 @@ cwlVersion: v1.0
 
 class: CommandLineTool
 
-label: MolGAN tool for generating small molecules 
+label: MolGAN tool for generating small molecules
 
 baseCommand: ["python", "/MolGAN/run_trained_model.py"]
 
 hints:
   DockerRequirement:
     dockerPull: ndonyapour/molgan
 
+# Set environment variables for the tool,
+# See: https://www.commonwl.org/user_guide/topics/environment-variables.html
+requirements:
+  EnvVarRequirement:
+    envDef:
+      RDKIT_ERROR_LOGGING: $(inputs.rdkit_error_logging)
+
 inputs:
   input_data_path:
     label: Path to the input data file
@@ -41,7 +48,7 @@ inputs:
     default: NP.gz
     inputBinding:
       prefix: --input_NP_Score_path
- 
+
   input_SA_Score_path:
     label: Output ceout file (AMBER ceout)
     doc: |-
@@ -60,10 +67,10 @@ inputs:
   input_model_dir:
     label: Input directory of trained models
     type: Directory
-    format: 
+    format:
     - edam:format_2330 # 'Textual format'
     inputBinding:
-      prefix: --input_model_dir 
+      prefix: --input_model_dir
     default: output
 
   output_log_path:
@@ -106,6 +113,16 @@ inputs:
       prefix: --num_samples
     default: 1000
 
+  rdkit_error_logging:
+    label: Enable or disable RDKit error logging
+    doc: |-
+      Enable or disable RDKit error logging
+    type: string?
+    format:
+    - edam:format_2330
+    # RDKit prints out all errors by default, which can pose issues for CI,
+    # particularly with large databases. It would be more efficient to suppress these errors.
+    default: "ON"
 outputs:
   output_log_path:
     label: Path to the log file
@@ -125,6 +142,13 @@ outputs:
       glob: $(inputs.output_sdf_path)
     format: edam:format_3814 # sdf
 
+  stderr:
+    type: File
+    outputBinding:
+      glob: stderr
+
+stderr: stderr
+
 
 $namespaces:
   edam: https://edamontology.org/

cwl_adapters/train_molgan_model.cwl

@@ -11,6 +11,13 @@ hints:
   DockerRequirement:
     dockerPull: ndonyapour/molgan
 
+# Set environment variables for the tool,
+# See: https://www.commonwl.org/user_guide/topics/environment-variables.html
+requirements:
+  EnvVarRequirement:
+    envDef:
+      RDKIT_ERROR_LOGGING: $(inputs.rdkit_error_logging)
+
 inputs:
   input_data_path:
     label: Path to the input data file
@@ -41,7 +48,7 @@ inputs:
     default: NP.gz
     inputBinding:
       prefix: --input_NP_Score_path
- 
+
   input_SA_Score_path:
     label: Output ceout file (AMBER ceout)
     doc: |-
@@ -74,10 +81,10 @@ inputs:
   output_model_dir:
     label: Output directory
     type: string
-    format: 
+    format:
     - edam:format_2330 # 'Textual format'
     inputBinding:
-      prefix: --output_model_dir 
+      prefix: --output_model_dir
     default: output
 
   validation_metrics:
@@ -86,13 +93,13 @@ inputs:
       The metrics are used during validation and testing
     type: string?
     format:
-    - edam:format_2330 
+    - edam:format_2330
     inputBinding:
       prefix: --validation_metrics
     default: 'np,logp,sas,qed,novelty,dc,unique,diversity,validity'
 
   num_epochs:
-    label: The number of training epochs 
+    label: The number of training epochs
     doc: |-
       The number of training epochs
       Type: int
@@ -111,7 +118,18 @@ inputs:
     format:
     - edam:format_2330
     inputBinding:
-      prefix: --save_frequency 
+      prefix: --save_frequency
+
+  rdkit_error_logging:
+    label: Enable or disable RDKit error logging
+    doc: |-
+      Enable or disable RDKit error logging
+    type: string?
+    format:
+    - edam:format_2330
+    # RDKit prints out all errors by default, which can pose issues for CI,
+    # particularly with large databases. It would be more efficient to suppress these errors.
+    default: "ON"
 
 outputs:
 
@@ -130,6 +148,13 @@ outputs:
       glob: $(inputs.output_model_dir)
     format: edam:format_2330 # 'Textual format
 
+  stderr:
+    type: File
+    outputBinding:
+      glob: stderr
+
+stderr: stderr
+
 $namespaces:
   edam: https://edamontology.org/
 

docker/dockerBuild.sh

@@ -25,7 +25,7 @@ sudo docker build --no-cache --pull -f Dockerfile_bash_scripts -t jakefennick/ba
 sudo docker build --no-cache --pull -f Dockerfile_calculate_net_charge -t jakefennick/calculate_net_charge .
 sudo docker build --no-cache --pull -f Dockerfile_mol2_to_pdbqt -t jakefennick/mol2_to_pdbqt .
 sudo docker build --no-cache --pull -f Dockerfile_nmr4md -t jakefennick/nmr4md .
-sudo docker build --no-cache --pull -f Dockerfile_openbabel -t jakefennick/openbabel .
+sudo docker build --no-cache --pull -f Dockerfile_openbabel -t ndonyapour/openbabel .
 sudo docker build --no-cache --pull -f Dockerfile_remove_terminal_residue_name_prefixes -t jakefennick/remove_terminal_residue_name_prefixes .
 sudo docker build --no-cache --pull -f Dockerfile_rename_residues_mol -t jakefennick/rename_residues_mol .
 sudo docker build --no-cache --pull -f Dockerfile_combine_structure -t ndonyapour/combine_structure .

dockerPull.sh

@@ -17,7 +17,7 @@ docker pull jakefennick/calculate_net_charge
 docker pull jakefennick/generate_conformers
 docker pull jakefennick/mol2_to_pdbqt
 docker pull jakefennick/nmr4md
-docker pull jakefennick/openbabel
+docker pull ndonyapour/openbabel
 docker pull jakefennick/remove_terminal_residue_name_prefixes
 docker pull jakefennick/rename_residues_mol
 docker pull ndonyapour/molgan

examples/scripts/Dockerfile_molgan

@@ -2,6 +2,9 @@
 FROM condaforge/mambaforge
 # NOT mambaforge-pypy3 (rdkit is incompatible with pypy)
 
+# RDKIT logging
+ENV RDKIT_ERROR_LOGGING="OFF"
+
 RUN apt-get update && apt-get install -y wget git
 
 # Clone MolGAN