diff options
| author | Hui Lan <lanhui@zjnu.edu.cn> | 2025-04-12 15:29:01 +0800 |
|---|---|---|
| committer | Hui Lan <lanhui@zjnu.edu.cn> | 2025-04-12 15:29:01 +0800 |
| commit | 9437f4637b10316765422ce30255b3aa2d1e267b (patch) | |
| tree | 79389f11db6f7f3ee93f64f0dd90164d48134917 | |
| parent | 933a7f22297873d114704d6ed0d34ac9eb012a0a (diff) | |
Save ENA records to XML files under Data/temp/
| -rw-r--r-- | Code/download_ena_records.py | 19 |
1 files changed, 17 insertions, 2 deletions
diff --git a/Code/download_ena_records.py b/Code/download_ena_records.py index 0f0cf30..b535c3e 100644 --- a/Code/download_ena_records.py +++ b/Code/download_ena_records.py @@ -1,10 +1,25 @@ -accession = 3702 -types = ['read_run', 'read_expriment', 'sample', 'study'] +# Download ENA records so that we can discover newly uploaded data +# 2025-04-12 +# Hui + +import os +import sys +import time +from configure import TEMP_DIR, UPDATE_NETWORK_LOG_FILE +from log import write_log_file +accession = 3702 # arabidopsis +types = ['read_run', 'read_expriment', 'sample', 'study'] for t in types: url = f'https://www.ebi.ac.uk/ena/browser/api/xml/links/taxon?accession={accession}&result={t}' print(url) + if not os.path.exists(TEMP_DIR): + sys.exit() + cmd = 'wget "%s" -O %s' % (url, os.path.join(TEMP_DIR, 'ena_'+t+'.xml')) # save XML data to file + os.system(cmd) + time.sleep(5) +write_log_file('ENA records updated. Check folder %s' % (TEMP_DIR), UPDATE_NETWORK_LOG_FILE) #https://www.ebi.ac.uk/ena/browser/api/xml/links/taxon?accession=3702&result=read_run #https://www.ebi.ac.uk/ena/browser/api/xml/links/taxon?accession=3702&result=read_experiment |