rdock-filter and enumerate_charges services working in basic form

Author: tdudgeon

data/nudt7/ligands.sdf.gz

@@ -14,7 +14,7 @@ CT_DY="application/x-squonk-service-descriptor-docker+yaml"
 CT_MM="multipart/mixed"
 
 
-for d in 'src/python/pipelines/dmpk' 'src/python/pipelines/docking' 'src/python/pipelines/rdkit'
+for d in 'src/python/pipelines/dmpk' 'src/python/pipelines/docking' 'src/python/pipelines/rdkit' 'src/python/pipelines/dimorphite'
 do
     for file in $d/*.dsd.yml
     do

data/nudt7/ligand.mol renamed to data/nudt7/refmol.mol

@@ -4,47 +4,48 @@
 * reference ligand.
 *
 * To test this manually run something like this:
-* nextflow -c src/nextflow/nextflow-docker.config run src/nextflow/docking/rdock-filter.nsd.nf --ligands data/nudt7/ligands.data.gz --ligand data/nudt7/ligand.mol --receptor data/nudt7/receptor.mol2 --num_dockings 5 -with-docker informaticsmatters/rdkit_pipelines
+* nextflow -c src/nextflow/nextflow-docker.config run src/nextflow/docking/rdock-filter.nsd.nf --ligands data/nudt7/ligands.data.gz --refmol data/nudt7/refmol.mol --receptor data/nudt7/receptor.mol2 --num_dockings 5
 */
 
-params.ligand = "$baseDir/ligand.mol"
+params.refmol = "$baseDir/refmol.mol"
 params.ligands = "$baseDir/ligands.data.gz"
-params.receptor = "$baseDir/receptor.mol2"
+params.receptor = "$baseDir/receptor.mol2.gz"
 params.chunk = 25
 params.num_dockings = 25
 params.top = 1
-params.score = null
-params.nscore = null
 params.limit = 0
 params.digits = 4
 params.threshold = 0.0
 params.field = 'SCORE.norm'
 
-ligand = file(params.ligand)
+refmol = file(params.refmol)
 ligands = file(params.ligands)
 receptor = file(params.receptor)
 
+expl2 = Channel.value( "Hello there $receptor" )
+
 process create_cavity {
 
     container 'informaticsmatters/rdock-mini:latest'
     beforeScript 'chmod g+w .'
 
     input:
-    file ligand
+    file refmol
     file receptor
 
     output:
     file 'receptor.prm' into prmfile
     file 'receptor.as' into asfile
 
     """
+    gunzip -c $receptor > receptor.mol2
     cat << EOF > receptor.prm
 RBT_PARAMETER_FILE_V1.00
-RECEPTOR_FILE $receptor
+RECEPTOR_FILE receptor.mol2
 RECEPTOR_FLEX 3.0
 SECTION MAPPER
     SITE_MAPPER RbtLigandSiteMapper
-    REF_MOL $ligand
+    REF_MOL $refmol
     RADIUS 3.0
     SMALL_SPHERE 1.0
     MIN_VOLUME 100
@@ -75,13 +76,14 @@ process dock_reference_ligand {
     file receptor
     file 'receptor.as' from asfile
     file 'receptor.prm' from prmfile
-    file ligand
+    file 'refmol.mol' from refmol
 
     output:
     file 'best_ligand.sdf' into best_ligand
 
     """
-    rbdock -i ligand.mol -r receptor.prm -p dock.prm -n $params.num_dockings -o docked_ligand > docked_ligand_out.log
+    gunzip -c $receptor > receptor.mol2
+    rbdock -i refmol.mol -r receptor.prm -p dock.prm -n $params.num_dockings -o docked_ligand > docked_ligand_out.log
     sdsort -n -s -fSCORE docked_ligand.sd | sdfilter -f'\$_COUNT <= 1' > best_ligand.sdf
     """
 }
@@ -124,13 +126,14 @@ process dock_ligands {
     file 'docked_part*.sd' into docked_parts
 
     """
+    gunzip -c $receptor > receptor.mol2
     rbdock -i $part -r receptor.prm -p dock.prm -n $params.num_dockings -o ${part.name.replace('ligands', 'docked')[0..-5]} > docked_out.log
     """
 }
 
 /* Filter, combine and publish the results.
 * Poses are only included if they are within ${params.threshold} of the best score obtained from docking the
-* target ligand into the same receptor (output of the dock_ligand process).
+* reference ligand into the same receptor (output of the dock_ligand process).
 */
 process combine_and_filter {
 

post-service-descriptors.sh

@@ -0,0 +1,87 @@
+---
+!<org.squonk.core.NextflowServiceDescriptor>
+serviceConfig:
+  id: pipelines.docking.rdock.filter.v1
+  "name": rDock filtering
+  "description": Dock ligands with rDock filtering poses relative to reference ligand
+  tags:
+  - rdock
+  - docking
+  - scoring
+  - 3d
+  - docker
+  - nextflow
+  resourceUrl: null
+  icon: icons/filter_molecules.png
+  inputDescriptors:
+  - name: receptor
+    mediaType: chemical/x-mol2
+    primaryType: org.squonk.types.Mol2File
+  - name: refmol
+    mediaType: chemical/x-mdl-molfile
+    primaryType: org.squonk.types.MolFile
+  - name: ligands
+    mediaType: application/x-squonk-dataset-molecule+json
+    primaryType: org.squonk.dataset.Dataset
+    secondaryType: org.squonk.types.MoleculeObject
+  outputDescriptors:
+  - name: output
+    mediaType: application/x-squonk-dataset-molecule+json
+    primaryType: org.squonk.dataset.Dataset
+    secondaryType: org.squonk.types.MoleculeObject
+  optionDescriptors:
+  - !<org.squonk.options.OptionDescriptor>
+    typeDescriptor: !<org.squonk.options.SimpleTypeDescriptor>
+      type: java.lang.Integer
+    key: arg.num_dockings
+    label: Number of dockings
+    description: Execute this many docking runs
+    defaultValue: 10
+    minValues: 1
+    maxValues: 1
+    visible: true
+    editable: true
+    modes:
+    - User
+  - !<org.squonk.options.OptionDescriptor>
+    typeDescriptor: !<org.squonk.options.SimpleTypeDescriptor>
+      type: java.lang.Integer
+    key: arg.top
+    label: Top scoring poses
+    description: Keep this many top scoring poses
+    defaultValue: 1
+    minValues: 1
+    maxValues: 1
+    visible: true
+    editable: true
+    modes:
+    - User
+  - !<org.squonk.options.OptionDescriptor>
+    typeDescriptor: !<org.squonk.options.SimpleTypeDescriptor>
+      type: java.lang.Float
+    key: arg.threshold
+    label: Score threshold
+    description: Filter out poses with relative scores above this value compared to reference
+    defaultValue: 0.0
+    minValues: 1
+    maxValues: 1
+    visible: true
+    editable: true
+    modes:
+    - User
+  executorClassName: org.squonk.execution.steps.impl.ThinDatasetNextflowInDockerExecutorStep
+thinDescriptors:
+  - input: ligands
+inputRoutes:
+- route: FILE
+- route: FILE
+- route: FILE
+outputRoutes:
+- route: FILE
+nextflowParams: |
+  params.receptor = 'receptor.mol2.gz'
+  params.refmol = 'refmol.mol'
+  params.ligands = 'ligands.data.gz'
+  ${binding.variables.containsKey('num_dockings') ? 'params.num_dockings = ' + num_dockings : ''}
+  ${binding.variables.containsKey('top') ? 'params.top = ' + top : ''}
+  ${binding.variables.containsKey('threshold') ? 'params.threshold = ' + threshold : ''}

src/nextflow/docking/rdock-filter.nsd.nf

@@ -0,0 +1,81 @@
+---
+"@class": org.squonk.core.DockerServiceDescriptor
+serviceConfig:
+  id: pipelines.dimorphite.enumerate_charges.v1
+  name: EnumerateCharges
+  description: Generate charge forms using Dimporphite_DL
+  tags:
+  - rdkit
+  - dimorphite
+  - charges
+  - enumerate
+  - docker
+  resourceUrl:
+  icon: icons/molecule_generator.png
+  inputDescriptors:
+  - primaryType: org.squonk.dataset.Dataset
+    secondaryType: org.squonk.types.MoleculeObject
+    mediaType: application/x-squonk-dataset-molecule+json
+    name: input
+  outputDescriptors:
+  - primaryType: org.squonk.dataset.Dataset
+    secondaryType: org.squonk.types.MoleculeObject
+    mediaType: application/x-squonk-dataset-molecule+json
+    name: output
+  optionDescriptors:
+  - "@class": org.squonk.options.OptionDescriptor
+    modes:
+    - User
+    typeDescriptor:
+      type: java.lang.Float
+      "@class": org.squonk.options.SimpleTypeDescriptor
+    editable: true
+    visible: true
+    description: Minimum pH
+    defaultValue: 5.0
+    label: Minimum pH value
+    key: arg.min_ph
+    minValues: 1
+    maxValues: 1
+  - "@class": org.squonk.options.OptionDescriptor
+    modes:
+      - User
+    typeDescriptor:
+      type: java.lang.Float
+      "@class": org.squonk.options.SimpleTypeDescriptor
+    editable: true
+    visible: true
+    description: Maximum pH
+    defaultValue: 9.0
+    label: Maximum pH value
+    key: arg.max_ph
+    minValues: 1
+    maxValues: 1
+  - modes:
+      - User
+    editable: true
+    "@class": org.squonk.options.OptionDescriptor
+    typeDescriptor:
+      type: java.lang.String
+      "@class": org.squonk.options.SimpleTypeDescriptor
+    key: arg.fragment_method
+    label: Fragment method
+    description: Approach to use for picking biggest molecular fragment
+    values:
+      - hac
+      - mw
+    defaultValue: hac
+    visible: true
+  executorClassName: org.squonk.execution.steps.impl.ThinDatasetDockerExecutorStep
+thinDescriptors:
+- input: input
+inputRoutes:
+- route: FILE
+outputRoutes:
+- route: FILE
+imageName: informaticsmatters/rdkit_pipelines
+command: >-
+  python -m pipelines.dimorphite.enumerate_charges -i ${PIN}input.data.gz -if json -o ${POUT}output -of json
+  ${binding.variables.containsKey('min_ph') ? ' --min-ph ' + min_ph : ''}
+  ${binding.variables.containsKey('max_ph') ? ' --max-ph ' + max_ph : ''}
+  ${binding.variables.containsKey('fragment_method') ? ' --fragment-method ' + fragment_method : ''}