[ensembl-dev] GTEx REST data

andrew126 at mac.com andrew126 at mac.com
Tue Feb 20 07:10:18 GMT 2018


Hi,

I am writing to inquire about the v6 GTEx eQTL data served by REST.

eQTL results in the original v6 GTEx downloads are of this format:
	gene: ENSG00000228327.2
	SNP: 1_1584002_A_G_b37
		(where, SNP is chr_position_ref_alt_assembly)
	beta: -0.167152
	t-test: -1.15677
	p-value: 0.252864

The beta is the eQTL effect relative to the alternate allele.  That is, in the above example, the G allele results in lower transcription of ENSG00000228327.

GTEx also released an rs# lookup for the v6 data:
	1       1584002 1_1584002_A_G_b37       A       G       1       rs6691927

The above data are returned by REST for Brain_Anterior_cingulate_cortex_BA24 via:
	https://rest.ensembl.org/eqtl/id/homo_sapiens/ENSG00000228327?content-type=application/json;tissue=Brain_Anterior_cingulate_cortex_BA24;variant_name=rs6691927

The returned data are:
	[{"statistic":"beta","value":-0.167152125808732},{"statistic":"p-value","minus_log10_p_value":0.597112188559852,"value":0.252864470319954}]

The data make no mention of the effect allele.  Is there a method by which the effect allele (alternate allele as declared in the GTEx analysis) can be determined and/or returned?

Direct interpretation seems complicated by 800k of the 11m v6 variants currently having more than 1 alternate allele in v90 Ensembl.
Another 140k of the variants have non-matching ref and alt alleles in the v90 assembly .. some appear to be simple strand flips, while others are less obvious .. for example, rs6691927 has changed from being an A/G variant to being a G/A variant (so one would need to know to grab the current alternate allele and change the sign of the beta value for correct interpretation).

Thank you for any guidance.

In addition, are there plans to update the eQTL data to v7?

Best,

Andrew






More information about the Dev mailing list