add scoring_onionnet workflow

cwl_adapters/autodock_vina.cwl

@@ -0,0 +1,221 @@
+
+#!/usr/bin/env cwl-runner
+cwlVersion: v1.1  # See `loadContents: true` below!
+
+class: CommandLineTool
+
+label: Wrapper of the AutoDock Vina software.
+
+doc: |-
+  This class performs docking of the ligand to a set of grids describing the target protein via the AutoDock Vina software.
+
+baseCommand: vina # NOTE: Only version >=1.2 supports --batch!
+arguments:
+# Need to parse box.pdb and pass in each number separately.
+# REMARK BOX CENTER:     0.102     0.019    -0.004 SIZE:    30.195    31.940    27.005
+# - "--dir" # Need to explicitly pass --dir . in --batch mode
+# - "."
+- "--center_x"
+- $(inputs.input_box_path.contents.split("\n")[0].split(" ").filter(function(s) {return !isNaN(parseFloat(s))})[0])
+- "--center_y"
+- $(inputs.input_box_path.contents.split("\n")[0].split(" ").filter(function(s) {return !isNaN(parseFloat(s))})[1])
+- "--center_z"
+- $(inputs.input_box_path.contents.split("\n")[0].split(" ").filter(function(s) {return !isNaN(parseFloat(s))})[2])
+- "--size_x"
+- $(inputs.input_box_path.contents.split("\n")[0].split(" ").filter(function(s) {return !isNaN(parseFloat(s))})[3])
+- "--size_y"
+- $(inputs.input_box_path.contents.split("\n")[0].split(" ").filter(function(s) {return !isNaN(parseFloat(s))})[4])
+- "--size_z"
+- $(inputs.input_box_path.contents.split("\n")[0].split(" ").filter(function(s) {return !isNaN(parseFloat(s))})[5])
+# NOTE: Cannot use a single javascript expression to create the entire arguments list because CWL treats it as a string:
+# "the `arguments` field is not valid because value is a str"
+#  ${
+#    var words = inputs.input_box_path.contents.split("\n")[0].split(" ");
+#    var nums = words.filter(function(s) {return !isNaN(parseFloat(s))});
+#    var args = {};
+#    args.push("--dir"); // Need to explicitly pass --dir . in --batch mode
+#    args.push(".");
+#    args.push("--center_x");
+#    args.push(nums[0]);
+#    args.push("--center_y");
+#    args.push(nums[1]);
+#    args.push("--center_z");
+#    args.push(nums[2]);
+#    args.push("--size_x");
+#    args.push(nums[3]);
+#    args.push("--size_y");
+#    args.push(nums[4]);
+#    args.push("--size_z");
+#    args.push(nums[5]);
+#    return args;
+#  }
+
+hints:
+  DockerRequirement:
+    dockerPull: jakefennick/autodock_vina
+
+requirements:
+  InlineJavascriptRequirement: {}
+
+inputs:
+  input_ligand_pdbqt_path:
+    label: Path to the input PDBQT ligand
+    doc: |-
+      Path to the input PDBQT ligand
+      Type: string
+      File type: input
+      Accepted formats: pdbqt
+      Example file: https://github.com/bioexcel/biobb_vs/raw/master/biobb_vs/test/data/vina/vina_ligand.pdbqt
+    type: File
+    format:
+    - edam:format_1476
+    inputBinding:
+      prefix: --ligand
+
+  input_receptor_pdbqt_path:
+    label: Path to the input PDBQT receptor
+    doc: |-
+      Path to the input PDBQT receptor
+      Type: string
+      File type: input
+      Accepted formats: pdbqt
+      Example file: https://github.com/bioexcel/biobb_vs/raw/master/biobb_vs/test/data/vina/vina_receptor.pdbqt
+    type: File
+    format:
+    - edam:format_1476
+    inputBinding:
+      #position: 2
+      prefix: --receptor
+
+  input_box_path:
+    label: Path to the PDB containing the residues belonging to the binding site
+    doc: |-
+      Path to the PDB containing the residues belonging to the binding site
+      Type: string
+      File type: input
+      Accepted formats: pdb
+      Example file: https://github.com/bioexcel/biobb_vs/raw/master/biobb_vs/test/data/vina/vina_box.pdb
+    type: File
+    format:
+    - edam:format_1476
+#    inputBinding:
+#      position: 3
+#      prefix: --input_box_path
+    loadContents: true  # requires cwlVersion: v1.1
+    # See https://www.commonwl.org/v1.1/CommandLineTool.html#Changelog
+    # Curiously, cwlVersion: v1.0 allows loadContents for outputs, but not inputs.
+
+  output_pdbqt_path:
+    label: Path to the output PDBQT file
+    doc: |-
+      Path to the output PDBQT file
+      Type: string
+      File type: output
+      Accepted formats: pdbqt
+      Example file: https://github.com/bioexcel/biobb_vs/raw/master/biobb_vs/test/reference/vina/ref_output_vina.pdbqt
+    type: string
+    format:
+    - edam:format_1476
+    inputBinding:
+      prefix: --out
+    default: system.pdbqt
+
+  output_log_path:
+    label: Path to the log file
+    doc: |-
+      Path to the log file
+      Type: string
+      File type: output
+      Accepted formats: log
+      Example file: https://github.com/bioexcel/biobb_vs/raw/master/biobb_vs/test/reference/vina/ref_output_vina.log
+    type: string
+    format:
+    - edam:format_2330
+    default: system.log
+
+  cpu:
+    label: The number of CPUs to use
+    doc: |-
+      The number of CPUs to use
+      Type: int
+    type: int?
+    format:
+    - edam:format_2330
+    inputBinding:
+      prefix: --cpu
+    default: 1
+
+  exhaustiveness:
+    label: exhaustiveness of the global search (roughly proportional to time).
+    doc: |-
+      exhaustiveness of the global search (roughly proportional to time).
+      Type: int
+    type: int?
+    format:
+    - edam:format_2330
+    inputBinding:
+      prefix: --exhaustiveness
+    default: 8
+
+  num_modes:
+    label: maximum number of binding modes to generate
+    doc: |-
+      Tmaximum number of binding modes to generate
+      Type: int
+    type: int?
+    format:
+    - edam:format_2330
+    inputBinding:
+      prefix: --num_modes
+    default: 9 
+
+  min_rmsd:
+    label: The minimum RMSD between output poses     
+    doc: |-
+      The minimum RMSD between output poses
+      Type: int
+    type: int?
+    format:
+    - edam:format_2330
+    inputBinding:
+      prefix: --min_rmsd
+    default: 1
+
+  energy_range:
+    label: maximum energy difference between the best binding mode and the worst one displayed (kcal/mol).
+    doc: |-
+      maximum energy difference between the best binding mode and the worst one displayed (kcal/mol).
+      Type: int
+    type: int?
+    format:
+    - edam:format_2330
+    inputBinding:
+      prefix: --energy_range
+    default: 3
+
+outputs:
+  output_pdbqt_path:
+    label: Path to the output PDBQT file
+    doc: |-
+      Path to the output PDBQT file
+    type: File
+    outputBinding:
+      glob: $(inputs.output_pdbqt_path)
+    format: edam:format_1476
+
+  output_log_path:
+    label: Path to the log file
+    doc: |-
+      Path to the log file
+    type: File
+    outputBinding:
+      glob: $(inputs.output_log_path)
+    format: edam:format_2330
+
+stdout: $(inputs.output_log_path)
+
+$namespaces:
+  edam: https://edamontology.org/
+
+$schemas:
+- https://raw.githubusercontent.com/edamontology/edamontology/master/EDAM_dev.owl

cwl_adapters/convert_pdb.cwl

@@ -0,0 +1,129 @@
+#!/usr/bin/env cwl-runner
+cwlVersion: v1.0
+
+class: CommandLineTool
+
+label: This class is a wrapper of the Open Babel tool.
+
+doc: |-
+  Small molecule format conversion for structures or trajectories. Open Babel is a chemical toolbox designed to speak the many languages of chemical data. It's an open, collaborative project allowing anyone to search, convert, analyze, or store data from molecular modeling, chemistry, solid-state materials, biochemistry, or related areas. Visit the official page.
+
+baseCommand: obabel
+#Usage:
+#obabel[-i<input-type>] <infilename> [-o<output-type>] -O<outfilename> [Options]
+#...
+#Options, other than -i -o -O -m, must come after the input files.
+arguments: [$(inputs.input_path), "-h", "-o", "pdb", "-O", $(inputs.output_pdb_path)]
+# NOTE: These arguments must be given individually; they cannot be concatenated together.
+# (e.g. -xrhn) Otherwise, all but the first argument will be silently ignored!
+
+# pdb format options
+# Read Options e.g. -as
+#   s  Output single bonds only
+#   b  Disable bonding entirely
+#   c  Ignore CONECT records
+
+# Write Options, e.g. -xo
+#   n  Do not write duplicate CONECT records to indicate bond order
+#   o  Write origin in space group label (CRYST1 section)
+
+hints:
+  DockerRequirement:
+    dockerPull: quay.io/biocontainers/biobb_chemistry:4.0.0--pyhdfd78af_1
+
+requirements:
+  InlineJavascriptRequirement: {}
+
+inputs:
+  first_molecule:
+    label: Index of the first molecule (1-based)
+    doc: |-
+      Input Index of the first molecule (1-based)
+      Type: string?
+    type: string?
+    format:
+    - edam:format_2330 # 'Textual format'
+    inputBinding:
+      prefix: -f
+
+  last_molecule:
+    label: Index of the last molecule (1-based)
+    doc: |-
+      Input Index of the last molecule (1-based)
+      Type: string?
+    type: string?
+    format:
+    - edam:format_2330 # 'Textual format'
+    inputBinding:
+      prefix: -l
+
+  input_path:
+    label: Path to the input file
+    doc: |-
+      Path to the input file
+      Type: string
+      File type: input
+      Accepted formats: dat, ent, fa, fasta, gro, inp, log, mcif, mdl, mmcif, mol, mol2, pdb, pdbqt, png, sdf, smi, smiles, txt, xml, xtc
+      Example file: https://github.com/bioexcel/biobb_chemistry/raw/master/biobb_chemistry/test/data/babel/babel.smi
+    type: File
+    format:
+    - edam:format_1637
+    - edam:format_1476
+    - edam:format_1929
+    - edam:format_1929
+    - edam:format_2033
+    - edam:format_3878
+    - edam:format_2030
+    - edam:format_1477
+    - edam:format_3815
+    - edam:format_1477
+    - edam:format_3815
+    - edam:format_3816
+    - edam:format_1476
+    - edam:format_1476
+    - edam:format_3603
+    - edam:format_3814
+    - edam:format_1196
+    - edam:format_1196
+    - edam:format_2033
+    - edam:format_2332
+    - edam:format_3875
+
+  output_pdb_path:
+    label: Path to the output file
+    doc: |-
+      Path to the output file
+      Type: string
+      File type: output
+      Accepted formats: pdb
+    type: string
+    format:
+    - edam:format_1476 # pdb
+    default: system.pdb
+
+  arg1:
+    label: Additional arguments
+    doc: |-
+      Additional arguments
+      Type: string?
+    type: string?
+    format:
+    - edam:format_2330 # 'Textual format'
+    inputBinding:
+      position: 1
+
+outputs:
+  output_pdb_path:
+    label: Path to the output file
+    doc: |-
+      Path to the output file
+    type: File
+    outputBinding:
+      glob: $(inputs.output_pdb_path)
+    format: edam:format_1476 # pdb
+
+$namespaces:
+  edam: https://edamontology.org/
+
+$schemas:
+- https://raw.githubusercontent.com/edamontology/edamontology/master/EDAM_dev.owl

cwl_adapters/onionnet_featurize.cwl

@@ -5,7 +5,7 @@ class: CommandLineTool
 
 label: Generate the features for the protein-ligand complexes using OnionNet V1
 
-baseCommand: ["python", "/onionnet/generate_features.py"]
+baseCommand: ["conda", "run", "-n", "py37", "python", "/onionnet/generate_features.py"]
 arguments: ["-inp", "input.dat"]
 
 hints:

cwl_adapters/onionnet_score.cwl

@@ -5,7 +5,7 @@ class: CommandLineTool
 
 label: OnionNet (version1) for rescoring of docking poses
 
-baseCommand: ["python", "/onionnet/predict.py"]
+baseCommand: ["conda", "run", "-n", "py37", "python", "/onionnet/predict.py"]
 
 hints:
   DockerRequirement: