Feature tagged by cgi as 'CNA', but biomarker_type set to 'nonsense'

[Brian]

@mayfielg can you take look at this? I think

[Georgia Mayfield]

Hey @bwalsh, I can't seem to replicate this behavior. There was an issue where no biomarker type was showing up, which I've fixed, but nowhere can I see the biomarker type set to 'nonsense'. Where are you seeing this?

[Alex H. Wagner, PhD]

This was from the CGI source, and the associated key that contained the response data was remapped from cgi to raw in the notebook that @bwalsh took screenshots from.

So for an association from CGI, you'd want to examine:
assocation['cgi']

[Brian]

I think, but have not yet confirmed, that it is being set here
https://github.com/ohsu-comp-bio/g2p-aggregator/blob/v0.9/harvester/feature_enricher.py#L113-L117

[Alex H. Wagner, PhD]

Another instance of this:

{'association': {'description': 'FLCN Everolimus (MTOR inhibitor) Responsive',
  'drug_labels': 'EVEROLIMUS',
  'environmentalContexts': [{'approved_countries': ['Canada', 'US'],
    'description': 'EVEROLIMUS',
    'id': 'CID6442177',
    'source': 'http://rdf.ncbi.nlm.nih.gov/pubchem/compound',
    'taxonomy': {'class': 'Macrolide lactams',
     'direct-parent': 'Macrolide lactams',
     'kingdom': 'Organic compounds',
     'superclass': 'Phenylpropanoids and polyketides'},
    'term': 'EVEROLIMUS',
    'toxicity': 'IC50 of 0.63 nM.',
    'usan_stem': 'immunosuppressives: immunosuppressant, rapamycin derivatives'}],
  'evidence': [{'description': 'Responsive',
    'evidenceType': {'sourceName': 'cgi'},
    'info': {'publications': ['http://www.ncbi.nlm.nih.gov/pubmed/25295501']}}],
  'evidence_label': 'C',
  'evidence_level': 3,
  'phenotype': {'description': 'thyroid cancer',
   'family': 'endocrine gland cancer',
   'type': {'id': 'DOID:1781',
    'source': 'http://purl.obolibrary.org/obo/doid',
    'term': 'thyroid cancer'}},
  'publication_url': 'http://www.ncbi.nlm.nih.gov/pubmed/25295501',
  'response_type': 'Responsive'},
 'dev_tags': [],
 'feature_names': 'FLCN deletion',
 'features': [{'alt': 'A',
   'biomarker_type': 'nonsense',
   'chromosome': '17',
   'description': 'FLCN deletion',
   'end': 17122384,
   'geneSymbol': 'FLCN',
   'links': ['http://myvariant.info/v1/variant/chr17:g.17122384G>A?assembly=hg19',
    'http://reg.genome.network/refseq/RS000041',
    'http://reg.genome.network/refseq/RS000017',
    'http://reg.genome.network/refseq/RS000765',
    'http://reg.genome.network/allele/CA498163528',
    'http://reg.genome.network/refseq/RS000065'],
   'name': 'FLCN deletion',
   'provenance': ['http://myvariant.info/v1/query?q=FLCN deletion',
    'http://reg.genome.network/allele?hgvs=NC_000017.10%3Ag.17122384G%3EA'],
   'provenance_rule': 'default_feature',
   'ref': 'G',
   'referenceName': 'GRCh37',
   'sequence_ontology': {'hierarchy': ['SO:0001060',
     'SO:0001537',
     'SO:0001878',
     'SO:0001564',
     'SO:0001576',
     'SO:0001791',
     'SO:0001580',
     'SO:0001818',
     'SO:0001650',
     'SO:0001992'],
    'name': 'stop_gained',
    'parent_name': 'sequence_variant',
    'parent_soid': 'SO:0001060',
    'soid': 'SO:0001587'},
   'start': 17122384,
   'synonyms': ['NG_008001.2:g.23119C>T',
    'NC_000017.11:g.17219070G>A',
    'LRG_325:g.23119C>T',
    'NC_000017.10:g.17122384G>A',
    'CM000679.1:g.17122384G>A',
    'chr17:g.17122384G>A',
    'CM000679.2:g.17219070G>A',
    'NC_000017.9:g.17063109G>A']}],
 'genes': ['FLCN'],
 'raw': {'Alteration': 'FLCN:del',
  'Alteration type': 'CNA',
  'Assay type': '',
  'Association': 'Responsive',
  'Biomarker': 'FLCN deletion',
  'Curator': 'RDientsmann',
  'Drug': 'Everolimus',
  'Drug family': 'MTOR inhibitor',
  'Drug full name': 'Everolimus (MTOR inhibitor)',
  'Drug status': '',
  'Evidence level': 'Case report',
  'Gene': 'FLCN',
  'Metastatic Tumor Type': '',
  'Primary Tumor type': 'Thyroid',
  'Source': 'PMID:25295501',
  'Targeting': '',
  'cDNA': [''],
  'gDNA': [''],
  'individual_mutation': [''],
  'info': [''],
  'region': [''],
  'strand': [''],
  'transcript': ['']},
 'source': 'cgi',
 'tags': []}

[Brian]

@ahwagner : can you double check? This data appears prior to v0.10's provenance rules.
The data I have for this association is:

$ cat cgi.json | grep "FLCN deletion" | jq .features
[
  {
    "provenance_rule": "is_deletion",
    "end": 17140502,
    "description": "FLCN deletion",
    "provenance": [
      "http://mygene.info/v3/query?q=FLCN&fields=genomic_pos_hg19"
    ],
    "start": 17115526,
    "referenceName": "GRCh37",
    "geneSymbol": "FLCN",
    "chromosome": "17",
    "name": "FLCN deletion"
  }
]

[Brian]

Regarding the original item reported at the top of this thread, I believe we have resolved this.
The current feature is:

{
  "provenance_rule": "is_deletion",
  "end": 1228428,
  "name": "STK11 deletion",
  "sequence_ontology": {
    "hierarchy": [
      "SO:0000110",
      "SO:0002072",
      "SO:0001059"
    ],
    "soid": "SO:0000159",
    "parent_name": "sequence_feature",
    "name": "deletion",
    "parent_soid": "SO:0000110"
  },
  "provenance": [
    "http://mygene.info/v3/query?q=STK11&fields=genomic_pos_hg19"
  ],
  "start": 1189406,
  "biomarker_type": "del",
  "referenceName": "GRCh37",
  "geneSymbol": "STK11",
  "chromosome": "19",
  "description": "STK11 deletion"
}

@ahwagner , @mayfielg : can we close this?

[Alex H. Wagner, PhD]

Will review today.