Ensembl ID conversion for a specific build

Question

[ChIP]

Hi,

I have ensembl gene Ids from GRCh37.70 and I guess I need to use ensembl genes 70. But unfortunately, I can only find this linkin archive that leads to ensembl genes 75. I think I need ensembl gene 70.

Could anyone help me with this.

Thank you

[GenoMax]

Currently available archives are 67, 74 and 75 so you would need to use of one of those. This table shows the various assemblies present in those versions.

Answer 1

[Denise CS]

Yes, IDs should be (and mostly are) stable across releases if the gene model itself has not changed. However if there are changes in the gene model (even if there are 'minor' and does not affect the coding bits e.g. UTRs were extended), a new ENS ID will be assigned.

As we are comparing the models across different versions of the assembly, you may find gene A with ENS ID A in release 70, which got ENS ID B in release 86 (the structure of gene A could have changed). Check this previous post for more details.

Have you thought in getting GTF from the the Ensembl release 70 available in the Ensembl FTP?

Archive older than 3 years old (plus a few exceptions) are not maintained for browser access but are available through the FTP.

[GenoMax]

Archive older than 3 years old (plus a few exceptions) are not maintained for browser access but are available through the FTP

That is good to know.

[i.sudbery]

Do gene IDs change when UTRs change? Surely the extent of a UTRs etc, is a transcript thing?

Also, when does the revision number get increased rather than the whole ID changing

e.g. When is ENSGXXXXXXXXX.1 changed to ENSGXXXXXXXXX.2 and when is it changed to ENSGYYYYYYYYY?

[ChIP]

Thank you so much.

The information and the link was very helpful.

Answer 2

[i.sudbery]

IDs should be stable across releases I think. That is if an ID exists in both releases, it should refer to the same thing. Thus if your ID exists in the latest ensembl, then it will reffer to the same gene or transcript (although the structure of that gene or transcript may have changed). If the ID does not exist in Ensembl 86, then that gene/transcript (not just the ID) has been retired, and you won't be able to find a corresponding ID however good your conversion is.