Merge remote-tracking branch 'origin/master'

Author: Alan Christie

src/nextflow/README.md

@@ -11,7 +11,7 @@ be found in the < project root >/src directory and images can be found on Docker
 Nextflow files names *.nsd.nf are destined for Squonk (and may have corresponding *.nsd.yml and *.nsd.config files).
 For normal use just create files like *.nf. The .nsd.yml extension is needed for it to be recognised by Squonk.
 
-## RDKit examples
+## RDKit implementations
 
 ### screen - a simple example
 
@@ -107,6 +107,23 @@ Launching `src/nextflow/docking/smog.nf` [cheesy_dijkstra] - revision: 220c72f4e
 [a6/06febe] Submitted process > results
 ```
 
+## rDock docking
+
+```
+$ nextflow run src/nextflow/docking/rdock.nf -c src/nextflow/docking/rdock.config -with-docker --ligands data/hivpr_ligprep_100.sdf.gz --protein data/hivpr_rdock.mol2 --asfile data/hivpr_rdock.as --prmfile data/hivpr_rdock.prm --num_dockings 10
+N E X T F L O W  ~  version 0.29.0
+Launching `src/nextflow/docking/rdock.nf` [festering_newton] - revision: e2ebfb5346
+[warm up] executor > local
+[51/31d685] Submitted process > sdsplit
+[db/e784c2] Submitted process > rdock (3)
+[aa/9927f7] Submitted process > rdock (1)
+[74/7b5c3b] Submitted process > rdock (2)
+[e1/c3301c] Submitted process > rdock (4)
+[bc/87e17d] Submitted process > results
+$
+```
+
+
 ## Usage in Squonk
 
 To be used in Squonk you must produce a `*.nsd.yml` file with the Squonk Service Descriptor. This has a corresponding Nextflow
@@ -129,7 +146,7 @@ in the service descriptor
 To test basic execution as it would take place in Squonk create a directory containing the input files that are needed by
 the pipeline (make sure they are named as expected by the pipeline) and then execute using the `execute` shell script that 
 would be generated by squonk. This file has a very simple content, something like this:
-```bash
+```
 #!/bin/sh
 nextflow run nextflow.nf -c nextflow.config <your-parameters> -with-docker
 ```
@@ -140,8 +157,8 @@ To explain this a bit:
 
 An example of how to execute would be this:
 
-```bash
+```
 docker run -it --rm -v $PWD:$PWD:z -w $PWD -v /var/run/docker.sock:/var/run/docker.sock informaticsmatters/nextflow sh -c 'nextflow run plip.nsd.nf -c plip.nsd.config --score 100.0 -with-docker'
 ``` 
 
-See [test-nextflow.sh](test-nextflow.sh) for some real examples.
+See [test-nextflow.sh](../..test-nextflow.sh) for some real examples.

src/nextflow/docking/plip.nf

@@ -51,20 +51,19 @@ process pli_scoring {
 */
 process results {
 
-	publishDir './', mode: 'copy'
 
 	input:
 	file ligands
 	file part from scored_parts.collect()
 
 	output:
-	file 'output.sdf.gz'
+	file 'output.sdf.gz' into results
 
 
 	"""
 	cat scored_part*.sdf | gzip > output.sdf.gz
 	"""
 }
 
-    
+results.println { "Results: $it" }
 

src/nextflow/docking/rdock.config

@@ -0,0 +1,7 @@
+docker.enabled = true
+// allow to run with SELinux
+docker.mountFlags = 'z'
+// this runs the containers as the current user
+docker.runOptions = '-u $(id -u):$(id -g)'
+// See here for information on this image: https://github.com/InformaticsMatters/rdock_docker
+process.container = 'informaticsmatters/rdock-mini:latest'

src/nextflow/docking/rdock.nf

@@ -0,0 +1,81 @@
+#!/usr/bin/env nextflow
+
+/* Example Nextflow pipeline that runs rDock docking
+*/
+
+
+params.ligands = 'ligands.sdf.gz'
+params.prmfile = 'receptor.prm'
+params.protein = 'receptor.mol2'
+params.asfile = 'receptor.as'
+params.chunk = 25
+
+params.num_dockings = 100
+params.top = 1
+params.score = null
+params.nscore = null
+params.limit = 0
+params.digits = 4
+
+
+ligands = file(params.ligands)
+protein = file(params.protein)
+prmfile = file(params.prmfile)
+asfile = file(params.asfile)
+
+/* Splits the input SD file into multiple files of ${params.chunk} records.
+* Each file is sent individually to the ligand_parts channel.
+* Parts are named so as to be in correct sorted area.
+*/
+process sdsplit {
+
+    container 'informaticsmatters/rdkit_pipelines:latest'
+
+	input:
+    file ligands
+
+    output:
+    file 'ligands_part*.sdf' into ligand_parts mode flatten
+    
+    
+    """
+    python -m pipelines_utils_rdkit.filter -i $ligands -c $params.chunk -d $params.digits -o ligands_part -of sdf --no-gzip
+    """
+}
+
+/* Docks each file from the ligand_parts channel sending each resulting SD file to the results channel
+*/
+process rdock {
+
+	input:
+    file part from ligand_parts
+	file protein
+	file prmfile
+	file asfile
+
+    output:
+    file 'docked_part*.sd' into docked_parts
+
+    """
+    rbdock -r $prmfile -p dock.prm -n $params.num_dockings -i $part -o ${part.name.replace('ligands', 'docked')[0..-5]} > docked_out.log
+    """
+}
+
+/* Filter, combine and publish the results
+*/
+process results {
+
+    publishDir './', mode: 'copy'
+
+	input:
+	file part from docked_parts.collect()
+
+	output:
+	file 'results.sdf' into results
+
+	"""
+	sdsort -n -s -fSCORE docked_part*.sd |${params.score == null ? '' : " sdfilter -f'\$SCORE <= $params.score' |"}${params.nscore == null ? '' : " sdfilter -f'\$SCORE.norm <= $params.nscore' |"} sdfilter -f'\$_COUNT <= ${params.top}' > results.sdf
+	"""
+}
+    
+results.println { "Results: $it" }

src/nextflow/docking/rdock.nsd.nf

@@ -66,7 +66,7 @@ process splitter {
 */
 process rdock {
 
-    container 'informaticsmatters/rdock'
+    container 'informaticsmatters/rdock-mini:latest'
     // change permissions on the work dir so that the rdock user in the container
     // can write to the directory that is owned by root
     beforeScript 'chmod g+w .'

src/nextflow/docking/smog.nf

@@ -47,24 +47,22 @@ process smog_scoring {
     """
  }
 
-/* Recombine and publish the results
+/* Recombine the results
 */
 process results {
 
-	publishDir './', mode: 'copy'
 
 	input:
 	file ligands
 	file part from scored_parts.collect()
 
 	output:
-	file 'output.sdf.gz'
+	file 'output.sdf.gz' into results
 
 
 	"""
 	cat scored_part*.sdf | gzip > output.sdf.gz
 	"""
 }
 
-    
-    
+results.println { "Results: $it" }

test-nextflow.sh

@@ -7,27 +7,21 @@ set -e
 echo 'Running screen+config'
 nextflow run src/nextflow/rdkit/screen+conformers.nf -c  src/nextflow/rdkit/screen.config -with-docker
 
-#echo 'Running screen in Squonk mode'
-#nextflow run src/nextflow/rdkit/smog.nf -c src/nextflow/docking/smog.config -with-docker --ligands data/smog/confs.sdf --protein data/smog/DCP2_1.pdb informaticsmatters/smog:latest
-
-echo 'Running RDock in squonk mode'
-sudo rm -rf tmp/*
-cp src/nextflow/docking/rdock.nsd.nf tmp/nextflow.nf
-cp src/nextflow/docking/rdock.nsd.config tmp/nextflow.config
-cp data/hivpr.config.zip tmp/config.zip
-cp data/dhfr_3d.data.gz tmp/ligands.data.gz
-cp data/dhfr_3d.metadata tmp/ligands.metadata
-cd tmp
-docker run -it --rm -v $PWD:$PWD:z -w $PWD -v /var/run/docker.sock:/var/run/docker.sock informaticsmatters/nextflow sh -c 'nextflow run nextflow.nf -c nextflow.config --num_dockings 1 --limit 40 --chunk 5 -with-docker'
-cd ..
-sudo rm -rf tmp/*
 
 echo 'Running SMoG2016 in basic mode'
 nextflow run src/nextflow/docking/smog.nf -c src/nextflow/docking/smog.config -with-docker --ligands data/smog/confs.sdf --protein data/smog/DCP2_1.pdb informaticsmatters/smog:latest
 
 echo 'Running PLI in basic mode'
 nextflow run src/nextflow/docking/plip.nf -c src/nextflow/docking/plip.config -with-docker --ligands data/smog/confs.sdf --protein data/smog/DCP2_1.pdb informaticsmatters/pli:latest
 
+echo 'Running rDock in basic mode'
+nextflow run src/nextflow/docking/rdock.nf -c src/nextflow/docking/rdock.config -with-docker\
+  --ligands data/hivpr_ligprep_100.sdf.gz\
+  --protein data/hivpr_rdock.mol2\
+  --asfile data/hivpr_rdock.as\
+  --prmfile data/hivpr_rdock.prm\
+  --num_dockings 2
+
 echo 'Running SMoG2016 in squonk mode'
 sudo rm -rf tmp/*
 cp src/nextflow/docking/smog.nsd.nf tmp/nextflow.nf
@@ -54,3 +48,14 @@ cd ..
 sudo rm -rf tmp/*
 
 
+echo 'Running rDock in squonk mode'
+sudo rm -rf tmp/*
+cp src/nextflow/docking/rdock.nsd.nf tmp/nextflow.nf
+cp src/nextflow/docking/rdock.nsd.config tmp/nextflow.config
+cp data/hivpr.config.zip tmp/config.zip
+cp data/dhfr_3d.data.gz tmp/ligands.data.gz
+cp data/dhfr_3d.metadata tmp/ligands.metadata
+cd tmp
+docker run -it --rm -v $PWD:$PWD:z -w $PWD -v /var/run/docker.sock:/var/run/docker.sock informaticsmatters/nextflow sh -c 'nextflow run nextflow.nf -c nextflow.config --num_dockings 1 --limit 40 --chunk 5 -with-docker'
+cd ..
+sudo rm -rf tmp/*