How to retrieve COSMIC variations using Ensembl REST API?

Question

[patrick.durand]

I'm failing to retrieve Cosmic variations for some human genes using Ensembl API.

Using a web browser with this link:

http://www.ensembl.org/Homo_sapiens/Transcript/ProtVariations?db=core;g=ENSG00000141510;r=17:7661779-7687550;t=ENST00000617185

illustrates that Cosmic variations are available for gene TP53 (ensembl id=ENSG00000141510); in the above URL, I actually retrieved variations for one on the many TP53 transcripts, ENST00000617185.

Ok, now, if I use one of these URLs (Ensembl API based) :

• using TP53 gene ID: curl -H 'Content-type:text/xml' -o variations.xml 'http://rest.ensembl.org/overlap/id/ENSG00000141510?feature=variation;variant_set=ph_variants'
• using TP53 transcript ID: curl -H 'Content-type:text/xml' -o variations.xml 'http://rest.ensembl.org/overlap/id/ENST00000617185?feature=variation;variant_set=ph_variants'

I can retrieve Clinvar, PhenCode, etc. mutations but no Cosmic ones.

Any idea of what I did wrong?

An optional question: to which kind of Ensembl objects (gene, transcript, protein) are attached the many variation features available from Ensembl? I could not find a documentation about that.

Answer 1

[Ben_Ensembl]

Hi Patrick,

We seperate somatic data from germline, so the query should be: curl -H 'Content-type:text/xml' 'http://rest.ensembl.org/overlap/id/ENSG00000141510?feature=somatic_variation'

<data id="COSM5493517" assembly_name="GRCh38" consequence_type="splice_region_variant" end="7662020" feature_type="somatic_variation" seq_region_name="17" source="COSMIC" start="766201 9" strand="1"> <alleles>COSMIC_MUTATION</alleles> </data> <data id="COSM18654" assembly_name="GRCh38" consequence_type="transcript_ablation" end="7676594" feature_type="somatic_variation" seq_region_name="17" source="COSMIC" start="7669609" s="" trand="1">

See: http://rest.ensembl.org/documentation/info/overlap_id for the list of feature types.

For the optional question: the genomic locations of variants are stored in our database, and are not linked to objects per se. The variant consequence(s)/effect(s) on gene(s) is inferred from the variant location in relation to the gene model.

Best wishes

Ben Ensembl Helpdesk

[patrick.durand]

Hi Ben,

ok, got it, thanks. By the way, do you have any documentation that gives more insights on how to combine feature types with variant_set. E.g. if I use "feature=variation", what are the appropriate values for "variant_set" ? I cannot find that information from the "overlap" service web page, neither from this page.

-Patrick

[Ben_Ensembl]

The 'variant_set' filter only works with the 'variation' feature type. We will change the text to clarify this for e!88, though.