[ensembl-dev] Possible error in grch37 database

[Wolf Beat]

Hi,


i'm using ensembl through both the REST API for grch37 and biomart on the old feb2014 archive (because i need a stable archive).


I do have a couple of unit tests to verify that my application works correctly, which is why i noticed one change i dont understand.


Until now, rs1054487 (G>A) had a MAF of 0.34, in ensembl as well as exac.

Now suddendly on grch37, the MAF changed to 0.0034.


This looks suspicously like a parsing error somewhere.


This must have been introduced quite recently. You can compare the grch37 and feb2014 versions here:

http://feb2014.archive.ensembl.org/Homo_sapiens/Variation/Explore?r=19:12771665-12772665;v=rs1054487;vdb=variation;vf=839082

http://grch37.ensembl.org/Homo_sapiens/Variation/Explore?r=19:12771665-12772665;v=rs1054487;vdb=variation;vf=757537


I know archives of grch37 are not planned. But i would like to repeat my desire for archives of feb2014. Not only to avoid issues like this to randomly appear, but also for a more stable API. There have been for example, to my knowledge, undocumented changes in the REST api for the VEP entry point. Also subtile changes in the biomart API have been made. Having stable versions would help to reduce those types of problems.


Kind regards


Beat Wolf