mods to screen nextflow workflows

Author: tdudgeon

src/nextflow/docking/rdock.nf

@@ -65,8 +65,6 @@ process rdock {
 */
 process results {
 
-    publishDir './', mode: 'copy'
-
 	input:
 	file part from docked_parts.collect()
 

src/nextflow/rdkit/screen-dataset.nsd.nf

@@ -14,7 +14,7 @@ target = file(params.input)
 
 process splitter {
 
-    container 'informaticsmatters/rdkit_pipelines'
+    container 'informaticsmatters/rdkit_pipelines:latest'
 
     input:
     file target
@@ -45,40 +45,24 @@ process rdkitScreen {
 
 process joiner {
 
-    container 'informaticsmatters/rdkit_pipelines'
+    container 'informaticsmatters/rdkit_pipelines:latest'
 
-    publishDir baseDir, pattern: "{output.data.gz,output.metadata}"
+    publishDir baseDir, mode: 'link'
 
     input:
+    file 'splitter_metrics.txt' from splitter_metrics
 	file parts from screened_parts.collect()
 
 	output:
-	file 'output_metrics.txt' into joiner_metrics
+	file 'output_metrics.txt'
 	file 'output.data.gz'
 	file 'output.metadata'
 
 	"""
 	zcat $parts | python -m pipelines_utils_rdkit.filter -if sdf -of json -o output --meta
-	"""
-}
-
-process metrics {
-
-    container 'informaticsmatters/rdkit_pipelines'
-
-    publishDir baseDir
-
-    input:
-    file 'splitter_metrics.txt' from splitter_metrics
-    file 'joiner_metrics.txt' from joiner_metrics
-
-    output:
-    file 'output_metrics.txt'
-
-    """
-    grep '__InputCount__' splitter_metrics.txt | sed s/__InputCount__/RDKitScreen/ > output_metrics.txt
+	mv output_metrics.txt joiner_metrics.txt
+	grep '__InputCount__' splitter_metrics.txt | sed s/__InputCount__/RDKitScreen/ > output_metrics.txt
     grep '__InputCount__' splitter_metrics.txt >> output_metrics.txt
     grep '__OutputCount__' joiner_metrics.txt >> output_metrics.txt
-    """
+	"""
 }
-

src/nextflow/rdkit/screen-multi-dataset.nsd.nf

@@ -15,7 +15,7 @@ query = file(params.query)
 
 process splitter {
 
-    container 'informaticsmatters/rdkit_pipelines'
+    container 'informaticsmatters/rdkit_pipelines:latest'
 
     input:
     file target
@@ -31,7 +31,7 @@ process splitter {
 
 process rdkitScreen {
 
-    container 'informaticsmatters/rdkit_pipelines'
+    container 'informaticsmatters/rdkit_pipelines:latest'
 
 	input:
     file part from target_parts
@@ -47,40 +47,24 @@ process rdkitScreen {
 
 process joiner {
 
-    container 'informaticsmatters/rdkit_pipelines'
+    container 'informaticsmatters/rdkit_pipelines:latest'
 
-    publishDir baseDir, pattern: "{output.data.gz,output.metadata}"
+    publishDir baseDir, mode: 'link'
 
     input:
+    file 'splitter_metrics.txt' from splitter_metrics
 	file parts from screened_parts.collect()
 
 	output:
-	file 'output_metrics.txt' into joiner_metrics
+	file 'output_metrics.txt'
 	file 'output.data.gz'
 	file 'output.metadata'
 
 	"""
 	zcat $parts | python -m pipelines_utils_rdkit.filter -if sdf -of json -o output --meta
-	"""
-}
-
-process metrics {
-
-    container 'informaticsmatters/rdkit_pipelines'
-
-    publishDir baseDir
-
-    input:
-    file 'splitter_metrics.txt' from splitter_metrics
-    file 'joiner_metrics.txt' from joiner_metrics
-
-    output:
-    file 'output_metrics.txt'
-
-    """
-    grep '__InputCount__' splitter_metrics.txt | sed s/__InputCount__/RDKitScreen/ > output_metrics.txt
+	mv output_metrics.txt joiner_metrics.txt
+	grep '__InputCount__' splitter_metrics.txt | sed s/__InputCount__/RDKitScreen/ > output_metrics.txt
     grep '__InputCount__' splitter_metrics.txt >> output_metrics.txt
     grep '__OutputCount__' joiner_metrics.txt >> output_metrics.txt
-    """
+	"""
 }
-

test-nextflow.sh

@@ -4,7 +4,7 @@
 
 set -e
 
-echo 'Running screen+config'
+echo 'Running screen+conformers in basic mode'
 nextflow run src/nextflow/rdkit/screen+conformers.nf -c  src/nextflow/rdkit/screen.config -with-docker
 
 
@@ -55,4 +55,26 @@ ln ../data/dhfr_3d.metadata ligands.metadata
 docker run -it --rm -v $PWD:$PWD:z -w $PWD -v /var/run/docker.sock:/var/run/docker.sock informaticsmatters/nextflow-docker:0.30.2 sh -c 'nextflow run nextflow.nf -c nextflow.config --num_dockings 1 --limit 40 --chunk 5 -with-docker'
 cd ..
 
+echo 'Running screen in squonk mode'
+sudo rm -rf tmp/*
+cd tmp
+ln ../src/nextflow/rdkit/screen-dataset.nsd.nf nextflow.nf
+ln ../src/nextflow/rdkit/screen-dataset.nsd.config nextflow.config
+ln ../data/dhfr_3d.data.gz input.data.gz
+ln ../data/dhfr_3d.metadata ligands.metadata
+docker run -it --rm -v $PWD:$PWD:z -w $PWD -v /var/run/docker.sock:/var/run/docker.sock informaticsmatters/nextflow-docker:0.30.2\
+  sh -c 'nextflow run nextflow.nf -c nextflow.config -with-docker --chunk 100 --simmin 0.5 --qsmiles "OC(=O)C1=CC=C(NC2=NC3=C(CN=C(C4=CC(Cl)=CC=C34)C3=C(F)C=CC=C3F)C=N2)C=C1"'
+cd ..
+
+echo 'Running screen-multi in squonk mode'
+sudo rm -rf tmp/*
+cd tmp
+ln ../src/nextflow/rdkit/screen-multi-dataset.nsd.nf nextflow.nf
+ln ../src/nextflow/rdkit/screen-multi-dataset.nsd.config nextflow.config
+ln ../data/dhfr_3d.data.gz target.data.gz
+ln ../data/nci100.data.gz query.data.gz
+docker run -it --rm -v $PWD:$PWD:z -w $PWD -v /var/run/docker.sock:/var/run/docker.sock informaticsmatters/nextflow-docker:0.30.2\
+  sh -c 'nextflow run nextflow.nf -c nextflow.config -with-docker --chunk 100 --simmin 0.55'
+cd ..
+
 sudo rm -rf tmp/*