new beta Monarch API does not serve inferred phenotypes, does serve EFO (Translator module compatibility issues)

[realmarcin]

The new Monarch beta API is not serving inferred phenotypes and added some EFO terms. The corresponding Monarch ticket is: https://github.com/monarch-initiative/dipper/issues/824

From Kent: "for HPO terms, there is some discussion on what we should be inferring for gene/variant to phenotype, by joining gene<-variant->disease->phenotype for now, we've removed all variant to phenotype data inferred in this way"

"for genes, we are only inferring disease to phenotype associations annotated with an obligate frequency qualifier"

"we've also removed all CTD gene to disease data"

Looks like going forward the best fix may be to replace the Monarch API call in ontobio with a direct SciGraph query?

Another (parallel) option is to expose past releases, or the missing inferred phenotypes, via the Monarch API (Melissa and Kent are aware).

SciGraph example via Kent -- this requires a HGNC CURIE I believe:

"one option would be to get the diseases associated with a gene (add the causal qualifier) and then get all phenotypes associated with that disease"

https://scigraph-data-dev.monarchinitiative.org/scigraph/cypher/execute.json?cypherQuery=MATCH%20(subject%3ANode{iri%3A%27HGNC%3A3582%27})%3C-[%3AGENO%3A0000418!*0..1]-(feature)-[%3ARO%3A0002200|RO%3A0003303|GENO%3A0000840|RO%3A0004011|RO%3A0004013|RO%3A0004016|RO%3A0004012|RO%3A0004014]-%3E(disease)-[relation%3ARO%3A0002200]-%3E(object%3Aphenotype)%20RETURN%20object.iri%20as%20phenotype&limit=1000000

In case its useful for testing, the previous phenotype data can be found here: https://archive.monarchinitiative.org/201902/tsv/gene_associations/gene_phenotype.9606.tsv

[realmarcin]

And a better diff example between beta and previous release -- 5 vs 46 phenotypes:

https://beta.monarchinitiative.org/gene/HGNC:3413#phenotype https://monarchinitiative.org/gene/HGNC:3413#phenotypes