aboutsummaryrefslogtreecommitdiff
path: root/workflows/pangenome-generate
diff options
context:
space:
mode:
authorAndreaGuarracino2020-09-26 12:12:17 +0200
committerAndreaGuarracino2020-09-26 12:12:17 +0200
commit7c12e4976337a063301be260cb3954bf4303f5e0 (patch)
treea25effc5a4afca5aee235e2b9e986abfab10ea12 /workflows/pangenome-generate
parent8e9247ab29db30e9bb7c87bef5f127b3b5cec699 (diff)
downloadbh20-seq-resource-7c12e4976337a063301be260cb3954bf4303f5e0.tar.gz
bh20-seq-resource-7c12e4976337a063301be260cb3954bf4303f5e0.tar.lz
bh20-seq-resource-7c12e4976337a063301be260cb3954bf4303f5e0.zip
script for processing the metadata of the ESR samples; moved delete_entries_on_arvados script in scripts directory
Diffstat (limited to 'workflows/pangenome-generate')
-rw-r--r--workflows/pangenome-generate/delete_entries_on_arvados.py34
1 files changed, 0 insertions, 34 deletions
diff --git a/workflows/pangenome-generate/delete_entries_on_arvados.py b/workflows/pangenome-generate/delete_entries_on_arvados.py
deleted file mode 100644
index d281456..0000000
--- a/workflows/pangenome-generate/delete_entries_on_arvados.py
+++ /dev/null
@@ -1,34 +0,0 @@
-import sys
-import arvados
-import arvados.collection
-
-from datetime import datetime
-
-date_time_str = '2020-08-20'
-date_time_obj = datetime.strptime(date_time_str, '%Y-%m-%d')
-
-api = arvados.api()
-keepclient = arvados.keep.KeepClient(api_client=api)
-
-validated = arvados.util.list_all(api.collections().list, filters=[
- ["owner_uuid", "=", sys.argv[1]],
-# ["properties.status", "=", "validated"]
-])
-
-# validated.sort(key=lambda v: v["portable_data_hash"])
-
-num_sample_deleted = 0
-for item in validated:
- sequence_label = item['properties']["sequence_label"]
-
- # The SRA samples start with SRR or ERR
- if not sequence_label.startswith('SRR') and not sequence_label.startswith('ERR'):
- created_at_obj = datetime.strptime(item["created_at"], '%Y-%m-%dT%H:%M:%S.%fZ')
- # print(item, created_at_obj)
-
- if created_at_obj < date_time_obj:
- api.collections().delete(uuid=item['current_version_uuid']).execute()
- num_sample_deleted += 1
- print(sequence_label)
-
-print('num_sample_deleted: {}'.format(num_sample_deleted))