diff options
author | Pjotr Prins | 2021-01-03 09:48:14 +0000 |
---|---|---|
committer | Pjotr Prins | 2021-01-03 09:48:14 +0000 |
commit | a10c61d1f5afca70f54b762beefb402372d7fd31 (patch) | |
tree | 40824a1aad7c532979c9dada50163d745d0e3373 /workflows/pull-data/genbank/ref.py | |
parent | a5ba1a8062e7116c2951762f86a6ae6d1638261d (diff) | |
download | bh20-seq-resource-a10c61d1f5afca70f54b762beefb402372d7fd31.tar.gz bh20-seq-resource-a10c61d1f5afca70f54b762beefb402372d7fd31.tar.lz bh20-seq-resource-a10c61d1f5afca70f54b762beefb402372d7fd31.zip |
genbank: technology parsing
Diffstat (limited to 'workflows/pull-data/genbank/ref.py')
-rw-r--r-- | workflows/pull-data/genbank/ref.py | 18 |
1 files changed, 0 insertions, 18 deletions
diff --git a/workflows/pull-data/genbank/ref.py b/workflows/pull-data/genbank/ref.py index 66c9fb0..d809d7f 100644 --- a/workflows/pull-data/genbank/ref.py +++ b/workflows/pull-data/genbank/ref.py @@ -1,23 +1,5 @@ # ---- BELOW IS JUST FOR REFERENCE ---- - # This script download and prepare data and metadata for assemblies samples - technology['assembly_method'] = 'http://purl.obolibrary.org/obo/GENEPIO_0001628' - - GBSeq_comment = GBSeq.find('GBSeq_comment') - if GBSeq_comment is not None and 'Assembly-Data' in GBSeq_comment.text: - prefix_split_string = '##Genome-Assembly' if GBSeq_comment.text.startswith('##Genome-') else '##Assembly' - - GBSeq_comment_text = GBSeq_comment.text.split( - '{}-Data-START## ; '.format(prefix_split_string) - )[1].split(' ; {}-Data-END##'.format(prefix_split_string))[0] - - for info_to_check, field_in_yaml in zip( - ['Assembly Method', 'Coverage', 'Sequencing Technology'], - ['alignment_protocol', 'sequencing_coverage', 'sample_sequencing_technology'] - ): - if info_to_check in GBSeq_comment_text: - tech_info_to_parse = GBSeq_comment_text.split('{} :: '.format(info_to_check))[1].split(' ;')[0] - if field_in_yaml == 'sequencing_coverage': # A regular expression would be better! try: |