Merge pull request PolusAI#243 from jfennick/wic_inline_input

add !ii inline input custom yaml tag

Author: sameeul

.vscode/settings.json

@@ -17,7 +17,9 @@
     "yaml.customTags": [
         "!&",
         "!*",
-        "!ii"
+        "!ii",
+        "!ii mapping",
+        "!ii sequence"
     ],
     "[yaml]": {
         "editor.suggest.showWords": false

examples/diffdock/run_diffdock.yml

@@ -6,9 +6,9 @@ steps:
         # "The query() method uses a slightly modified Python syntax by default.
         # For example, the & and | (bitwise) operators have the precedence of their boolean cousins, and and or.
         # This is syntactically valid Python, however the semantics are different."
-        query: '(Kd_Ki == "Kd") and (value < 0.001)'
-        max_row: 1 #25 # Use 1 for CI
-        convert_Kd_dG: True
+        query: !ii '(Kd_Ki == "Kd") and (value < 0.001)'
+        max_row: !ii 1 #25 # Use 1 for CI
+        convert_Kd_dG: !ii True
       out:
       - output_pdb_paths: !& pdbbind_pdbs
       - output_sdf_paths: !& pdbbind_sdfs
@@ -56,17 +56,17 @@ steps:
       in:
         protein_path: !* final_pdbbind_pdbs.pdb
         ligand_path: !* final_sanitized_sdfs
-        samples_per_complex: 20 # figure 3 left in DiffDock paper
-        inference_steps: 20 # figure S11 in DiffDock paper
+        samples_per_complex: !ii 20 # figure 3 left in DiffDock paper
+        inference_steps: !ii 20 # figure S11 in DiffDock paper
       out:
       - output_files: !& diffdock_poses
 
   - rank_diffdock_poses:
       scatter: [diffdock_poses]
       in:
-        top_n_confident: 1000 # if only using top_percent_confidence, then set top_n_confident to trivially high number
+        top_n_confident: !ii 1000 # if only using top_percent_confidence, then set top_n_confident to trivially high number
         # if only want to use top_n_confident, then set top_percent_confidence to 100
-        top_percent_confidence: 33 # take top third of most confident poses, see figure 3 right in DiffDock paper
+        top_percent_confidence: !ii 33 # take top third of most confident poses, see figure 3 right in DiffDock paper
         diffdock_poses: !* diffdock_poses
       out:
       - output_poses: !& output_poses
@@ -75,7 +75,7 @@ steps:
       scatter: [predicted_poses]
       in:
         predicted_poses: !* output_poses
-        centroid_cutoff: 5 # if centroid of all poses are within cutoff then only keep most confident pose, requires visual inspection
+        centroid_cutoff: !ii 5 # if centroid of all poses are within cutoff then only keep most confident pose, requires visual inspection
 
 wic:
   graphviz:

examples/diffdock/run_diffdock_weekly.yml

@@ -8,4 +8,4 @@ wic:
         steps:
           (1, extract_pdbbind_refined):
             in:
-              max_row: 16 # Override default 1
+              max_row: !ii 16 # Override default 1

examples/docking/align_protein_CA_pymol.yml

@@ -16,15 +16,15 @@ steps:
         input_index_path: !* MOL.ndx
         # Which causes the first column of this file to be all zeros
         # (which ruins the timeseries plot, but not the histogram)
-        output_xvg_path: rmsd_equil_ligand_notime.txt # Use .txt to avoid plotting
+        output_xvg_path: !ii rmsd_equil_ligand_notime.txt # Use .txt to avoid plotting
       out:
       - output_xvg_path: !& rmsd_equil_ligand_notime
   - bash_xvg:
       in:
-        script: /replace_first_column.sh # NOTE: Initial / required
+        script: !ii /replace_first_column.sh # NOTE: Initial / required
         # But we can recover the time column by copying it from another xvg file.
         input_xvg1_path: !* rmsd_equil_ligand_notime
         input_xvg2_path: !* rmsd_equil_mainchain.xvg
-        output_xvg_path: rmsd_equil_ligand_no_fit.xvg
+        output_xvg_path: !ii rmsd_equil_ligand_no_fit.xvg
       out:
       - output_xvg_path: !& rmsd_equil_ligand_no_fit.xvg

examples/docking/assign_partial_charges.yml

@@ -8,7 +8,7 @@ inputs:
 steps:
 - convert_mol2:
     in:
-      input_path: ~input_path #!* ligand_min.sdf
+      input_path: input_path #!* ligand_min.sdf
 #   out:
 #   - output_mol2_path: !& ligand_min.mol2
 # NOTE: If we directly convert from sdf to pdbqt, openbabel will NOT add

examples/docking/autodock_vina_rescore.yml

@@ -9,25 +9,25 @@ inputs:
 steps:
 - python_script:
     in:
-      script: ../scripts/atomselect.py
-      dockerPull: jakefennick/atomselect
-      selection_string: protein # Extract the protein from the last timestep
-      input_pdb_path: ~input_pdb_path # !* prod.pdb
+      script: !ii ../scripts/atomselect.py
+      dockerPull: !ii jakefennick/atomselect
+      selection_string: !ii protein # Extract the protein from the last timestep
+      input_pdb_path: input_pdb_path # !* prod.pdb
 # Assign partial charges (protein)
 # NOTE: Although we only start off with one protein, the final coordinates
 # after doing MD with different ligands will be different; hence scatter
 - convert_mol2:
 - convert_pdbqt:
     in:
-      arg1: -xr # Receptor needs to be rigid
+      arg1: !ii -xr # Receptor needs to be rigid
     out:
     - output_pdb_path: !& protein_prod.pdbqt
 - python_script:
     in:
-      script: ../scripts/atomselect.py
-      dockerPull: jakefennick/atomselect
-      selection_string: resname MOL # Extract the ligand from the last timestep
-      input_pdb_path: ~input_pdb_path # !* prod.pdb
+      script: !ii ../scripts/atomselect.py
+      dockerPull: !ii jakefennick/atomselect
+      selection_string: !ii resname MOL # Extract the ligand from the last timestep
+      input_pdb_path: input_pdb_path # !* prod.pdb
     # out:
     # - output_pdb_path: !& ligand_temp.pdbqt
 
@@ -41,7 +41,7 @@ steps:
 # ATOM     53   BR MOL Z   1     -14.786 -32.819  20.524  1.00  0.00          BR
 - fix_pdb_atom_column:
     in:
-      input_helper_structure_path: ~input_ligand_pdb_path
+      input_helper_structure_path: input_ligand_pdb_path
 # Assign partial charges (ligand)
 - convert_mol2:
 - convert_pdbqt:
@@ -51,8 +51,8 @@ steps:
     in:
       input_ligand_pdbqt_path: !* ligand_prod.pdbqt
       input_receptor_pdbqt_path: !* protein_prod.pdbqt
-      score_only: True
-      #local_only: True
+      score_only: !ii True
+      #local_only: !ii True
     out:
     - output_log_path: !& vina_rescore.log
     - docking_score: !& docking_rescores

examples/docking/convert_ligand_mol2_to_pdbqt_mdanalysis.yml

@@ -1,14 +1,14 @@
 steps:
   - python3_mol2_to_pdbqt:
       in:
-        script: /mol2_to_pdbqt.py # NOTE: Initial / required
+        script: !ii /mol2_to_pdbqt.py # NOTE: Initial / required
         input_mol2_path: !* conformer.mol2
       # out:
       # - output_pdb_path: !& ligand_periods.pdbqt
   - bash_pdb:
       in:
         # MDAnalysis adds a . to the final atomtype column, which causes autodock to crash.
-        script: /pdbqt_remove_trailing_period.sh # NOTE: Initial / required
+        script: !ii /pdbqt_remove_trailing_period.sh # NOTE: Initial / required
 #        input_pdb_path: !* ligand_periods.pdbqt
 #      out:
 #      - output_pdb_path: !& ligand_keywords.pdbqt
@@ -18,7 +18,7 @@ steps:
       # Even if there is actually only one model, autodock simply looks for MODEL and ENDMDL.
       # Unfortunately, all of the errors crash with the same cryptic message:
       # Parse error on line ... in file ".../ligand.pdbqt": Unknown or inappropriate tag
-        script: /pdbqt_remove_keywords.sh # NOTE: Initial / required
+        script: !ii /pdbqt_remove_keywords.sh # NOTE: Initial / required
 #        input_pdb_path: !* ligand_keywords.pdbqt
 #      out:
 #      - output_pdb_path: !& ligand_temp.pdbqt
@@ -27,7 +27,7 @@ steps:
       # Add ROOT, ENDROOT, TORSDOF keywords (if necessary).
       # Unfortunately, all of the errors crash with the same cryptic message:
       # Parse error on line ... in file ".../ligand.pdbqt": Unknown or inappropriate tag
-        script: /pdbqt_remove_flex.sh # NOTE: Initial / required
+        script: !ii /pdbqt_remove_flex.sh # NOTE: Initial / required
 #        input_pdb_path: !* ligand_temp.pdbqt
       out:
       - output_pdb_path: !& ligand_rigid.pdbqt

examples/docking/convert_ligand_mol2_to_pdbqt_obabel.yml

@@ -26,19 +26,19 @@ steps:
       # Even if there is actually only one model, autodock simply looks for MODEL and ENDMDL.
       # Unfortunately, all of the errors crash with the same cryptic message:
       # Parse error on line ... in file ".../ligand.pdbqt": Unknown or inappropriate tag
-        script: /pdbqt_remove_keywords.sh # NOTE: Initial / required
+        script: !ii /pdbqt_remove_keywords.sh # NOTE: Initial / required
 #        input_pdb_path: !* ligand_keywords.pdbqt
-        output_pdb_path: ligand_flex.pdbqt
+        output_pdb_path: !ii ligand_flex.pdbqt
       out:
       - output_pdb_path: !& ligand_flex.pdbqt
   - bash_pdb:
       in:
       # Add ROOT, ENDROOT, TORSDOF keywords (if necessary).
       # Unfortunately, all of the errors crash with the same cryptic message:
       # Parse error on line ... in file ".../ligand.pdbqt": Unknown or inappropriate tag
-        script: /pdbqt_remove_flex.sh # NOTE: Initial / required
+        script: !ii /pdbqt_remove_flex.sh # NOTE: Initial / required
         input_pdb_path: !* ligand_flex.pdbqt
-        output_pdb_path: ligand_rigid.pdbqt
+        output_pdb_path: !ii ligand_rigid.pdbqt
       out:
       - output_pdb_path: !& ligand_rigid.pdbqt
 

examples/docking/convert_pdbqt.yml

@@ -11,19 +11,19 @@ inputs:
 steps:
 - convert_mol2:
     in:
-      input_path: ~input_sdf_path
+      input_path: input_sdf_path
 
 - convert_pdbqt:
     out:
     - output_pdb_path: !& mol_prod.pdbqt
 
 - convert_mol2:
     in:
-      input_path: ~input_pdb_path
+      input_path: input_pdb_path
 
 - convert_pdbqt:
     in:
-      arg1: -xr # Receptor needs to be rigid
+      arg1: !ii -xr # Receptor needs to be rigid
     out:
     - output_pdb_path: !& receptor_prod.pdbqt
 wic:

examples/docking/docking.yml

@@ -11,7 +11,7 @@ steps:
 # to weaken the binding free energy! (as reported by autodock vina)
   - flc.yml:
       in:
-        sdf_path: ~sdf_path
+        sdf_path: sdf_path
 #  - minimize_ligand_only.yml:
 
 # NOTE: We converted to mol2 format above because it allows explicit charges.
@@ -24,7 +24,7 @@ steps:
 # "ERROR: more than one residue detected '{'UNL', 'MOL'}'"
   - python3_mol2_to_mol2:
       in:
-        script: /rename_residues_mol.py # NOTE: Initial / required
+        script: !ii /rename_residues_mol.py # NOTE: Initial / required
         input_mol2_path: !* ligand_min.mol2
       out:
       - output_mol2_path: !& conformer.mol2
@@ -51,7 +51,7 @@ steps:
 # NOTE: acpype doesn't add forcefield or water topology #include lines.
   - bash_top:
       in:
-        script: /gmx_add_topology_includes.sh # NOTE: Initial / required
+        script: !ii /gmx_add_topology_includes.sh # NOTE: Initial / required
         input_top_path: !* ligand_GMX.top
       out:
       - output_top_path: !& ligand_GMX_includes.top
@@ -78,15 +78,15 @@ steps:
   - extract_model_pdbqt:
       in:
         input_pdbqt_path: !* poses_ligand.pdbqt
-        output_pdbqt_path: pose_ligand.pdbqt
-        config:
+        output_pdbqt_path: !ii pose_ligand.pdbqt
+        config: !ii
           model: 1 # NOTE: score, rmsd l.b., rmsd u.b. stored in REMARK lines
       out:
       - output_pdbqt_path: !& pose_ligand.pdbqt
   - convert_xyz:
       in:
         input_path: !* pose_ligand.pdbqt
-        output_xyz_path: pose_ligand.xyz
+        output_xyz_path: !ii pose_ligand.xyz
       out:
       - output_xyz_path: !& pose_ligand.xyz
 # Molecular Dynamics combine receptor & ligand