diff options
author | Peter Amstutz | 2020-11-10 11:52:37 -0500 |
---|---|---|
committer | Peter Amstutz | 2020-11-10 11:52:37 -0500 |
commit | c01188ec20936462357b317f81567aadc64c8f33 (patch) | |
tree | ec01a8b2c5fa8fe87b0c750dad28f6a7f1a1a028 /workflows/pangenome-generate | |
parent | 74d46196ae69c0c557f64593910ada48d84b3654 (diff) | |
download | bh20-seq-resource-uuid-for-resource.tar.gz bh20-seq-resource-uuid-for-resource.tar.lz bh20-seq-resource-uuid-for-resource.zip |
Use arvados uuids for RDF subjects.uuid-for-resource
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Diffstat (limited to 'workflows/pangenome-generate')
-rw-r--r-- | workflows/pangenome-generate/collect-seqs.py | 5 |
1 files changed, 4 insertions, 1 deletions
diff --git a/workflows/pangenome-generate/collect-seqs.py b/workflows/pangenome-generate/collect-seqs.py index 1a0807c..225a61f 100644 --- a/workflows/pangenome-generate/collect-seqs.py +++ b/workflows/pangenome-generate/collect-seqs.py @@ -36,11 +36,14 @@ if len(sys.argv) > 3: for item in validated: pdh = item["portable_data_hash"] + uuid = item["uuid"] with arvados.collection.CollectionReader(pdh, api_client=api, keep_client=keepclient) as col: with col.open("sequence.fasta", "rt") as fa: - subject = "http://covid19.genenetwork.org/resource/%s" % pdh + subject = "http://covid19.genenetwork.org/resource/%s" % uuid label = fa.readline().strip() merged_metadata.write("<%s> <http://biohackathon.org/bh20-seq-schema/original_fasta_label> \"%s\" .\n" % (subject, label[1:].replace('"', '\\"'))) + merged_metadata.write("<%s> <http://biohackathon.org/bh20-seq-schema/collection_pdh> \"%s\" .\n" % (subject, pdh)) + merged_metadata.write("<%s> <http://biohackathon.org/bh20-seq-schema/collection_version> \"%s\" .\n" % (subject, item["version"])) skip = (subject in blacklist or label[1:] in blacklist) if skip: merged_metadata.write("<%s> <http://biohackathon.org/bh20-seq-schema/excluded_from_graph> \"true\"^^<http://www.w3.org/2001/XMLSchema#boolean> .\n" % subject) |