Extract subset of the all-sequences fasta by running a sparql query.

Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>

2 files changed, 31 insertions, 0 deletions

diff --git a/workflows/pangenome-generate/from_sparql.cwl b/workflows/pangenome-generate/from_sparql.cwl
new file mode 100644
index 0000000..5bc0792
--- /dev/null
+++ b/workflows/pangenome-generate/from_sparql.cwl
@@ -0,0 +1,23 @@
+cwlVersion: v1.1
+class: CommandLineTool
+requirements:
+  DockerRequirement:
+    dockerFile: |
+      FROM debian:10
+      RUN apt-get update && apt-get -yq --no-install-recommends install samtools python3-rdflib
+    dockerImageId: rdflib-and-samtools
+inputs:
+  script:
+    type: File
+    default:
+      class: File
+      location: from_sparql.py
+  metadata: File
+  fasta:
+    type: File
+    secondaryFiles: [.fai]
+  query: string
+stdout: selected.fasta
+outputs:
+  selected: stdout
+arguments: [python3, $(inputs.script), $(inputs.metadata), $(inputs.fasta), $(inputs.query)]
diff --git a/workflows/pangenome-generate/from_sparql.py b/workflows/pangenome-generate/from_sparql.py
new file mode 100644
index 0000000..4610cad
--- /dev/null
+++ b/workflows/pangenome-generate/from_sparql.py
@@ -0,0 +1,8 @@
+from rdflib import Graph
+import sys
+import subprocess
+g = Graph()
+g.parse(sys.argv[1], format="nt")
+res = g.query(sys.argv[3])
+for r in res:
+    subprocess.run(["samtools", "faidx", sys.argv[2], r[0]])