griffithlab / civic-api-client Goto Github PK

Each variant should have a representative transcript. The ID for this transcript should have the following format:

ENST00000355413.4

Hi @LeiChen0218. Can we change the behavior of civic-api-client variants-list so that it will not report a variant unless it has at least one accepted evidence item?

If a gene has one or more variants, and these variants have at least X total evidence statements, display the gene as ready for a gene summary to be written (default value of X could be 10).

Right now the api has a /genes endpoint that returns a subset of user-defined length with all genes and corresponding variants, variant groups, descriptions, etc. for those genes. It would be nice if this endpoint instead returned a list of ALL genes in the database, limited only to the id, name, and entrez_id fields, so that one could use this as a quick reference for getting the appropriate CIViC identifiers for detailed queries (e.g. if I'm interested in BRAF I could use the returned result {"id":5,"name":"BRAF","entrez_id":673} to do a subsequent API call for BRAF at https://civic.genome.wustl.edu/api/genes/5)

Each variant that has a representative transcript should specify the ensembl database version it came from. This version should be indicated as a simple integer. (e.g. '75')

Obi's suggestion:
Creates a tsv file of all variants in CIViC with representative transcript and coordinates available.
Outputs might look something like:
chr start stop ref var rep_transcript drugs diseases summary number_evidence_statements url_for_civic_variant

Scan for variants where:

• start and end are defined
• (end - start) > 5 bases
• ref and/or var base are not defined

For example,

• If an evidence item is associated with a variant that has EXPRESSION in the name, the variant origin value should be N/A. Any other value is likely wrong.

Sometimes I choose the wrong options when running this tool and want to rerun it a different way. When I do a I get a full traceback. I assume it is possible to catch this and exit gracefully right?

Something like the first answer here perhaps:
http://stackoverflow.com/questions/21120947/catching-keyboardinterrupt-in-python-during-program-shutdown

It would be awesome to add functionality for evidence level filtering so that we could (for example) get all evidence IDs without a proper DOID.

Find variants with at least X evidence statements, but where a variant summary has not been created (or it is still short, e.g. < Y characters).

Default value of X, minimum evidence statements could be 5. Default value of Y, minimum length of the variant summary could be 50 characters.

Hey guys,

I am getting an error when I try to run this now:

civic-api-client variants-list --no-coords --max-gene-count 10

Max number of genes to query is  10
Traceback (most recent call last):
  File "/Users/mgriffit/git/civic/civic-api-client/myenv/bin/civic-api-client", line 11, in <module>
    sys.exit(main())
  File "/Users/mgriffit/git/civic/civic-api-client/myenv/lib/python2.7/site-packages/civic_api_client/command_line.py", line 21, in main
    return vl1.main()
  File "/Users/mgriffit/git/civic/civic-api-client/myenv/lib/python2.7/site-packages/civic_api_client/variants_lister.py", line 249, in main
    self.create_filtered_variants_list()
  File "/Users/mgriffit/git/civic/civic-api-client/myenv/lib/python2.7/site-packages/civic_api_client/variants_lister.py", line 244, in create_filtered_variants_list
    self.filter_variants()
  File "/Users/mgriffit/git/civic/civic-api-client/myenv/lib/python2.7/site-packages/civic_api_client/variants_lister.py", line 210, in filter_variants
    vd1 = VariantDetails(self.args, variant_details)
  File "/Users/mgriffit/git/civic/civic-api-client/myenv/lib/python2.7/site-packages/civic_api_client/variants_lister.py", line 25, in __init__
    self.parse_variant_details(variant_details)
  File "/Users/mgriffit/git/civic/civic-api-client/myenv/lib/python2.7/site-packages/civic_api_client/variants_lister.py", line 47, in parse_variant_details
    self.gene_name = variants['entrez_name']
NameError: global name 'variants' is not defined

Many initial studies (especially phase I clinical trials) occur before the field has settled on a drug name. For example "AZD9291" became "osimertinib". It is a curation task to migrate towards the standardized and most widely accepted generic name for every drug. We should come up with a way to scan for use of such synonyms that could be replaced with the generic name according to some canonical source.

For example, the NCI Drug Dictionary:

http://www.cancer.gov/publications/dictionaries/cancer-drug