Can APPRIS be used to determine principal isoforms in older Ensembl releases?

Question

[alex.rubinsteyn]

I'm working with variants which were aligned against an older genome (hg19) and, thus,annotated using an older Ensembl release (75). Would it still make sense to use APPRIS to determine the principal isoform for each coding variant? Are Ensembl IDs stable across releases (in the sense that a particular transcript ID always refers to the same set of exons)?

Answer 1

[Denise CS]

Ensembl IDs are stable across releases and even assemblies: so a given ENST will and should refer to the same set of exons/introns. Release 77 was the first time Ensembl showed the data from APPRIS but this annotation system was published more than a year before.

[alex.rubinsteyn]

Thanks Denise,

Do you think the principal transcripts in APPRIS would still exist in e.g. Ensembl release 54?

[Denise CS]

I believe there were fewer spliced isoforms annotated back them in NCBI36. If I take one example BRCA2 I can see that we now have 7 transcripts and back then we had 1. For DMD, there were 15 isoforms in 2009 and 30 transcripts now. The stable IDs then and now should be the same. So what is annotated as PIs now can be traced back to the annotation of NCBI36. APPRIS was originally developed using Ensembl release 56. Would it be worth contacting APPRIS developers? Failing having APPRIS for Ensembl release 54, I'd focus on the transcripts that have got CCDS and failed that I'd focus on those resulting from the merged set between Ensembl and HAVANA.

Answer 2

[appris.cnio]

Hi Alex,

Can APPRIS be used to determine principal isoforms in older Ensembl releases?

In the website of APPRIS you can find the "Principal Isoforms" for hg19 (Ensembl 74-75). And for more species.

• http://appris.bioinfo.cnio.es
• http://appris.bioinfo.cnio.es/#/downloads (dowload section).

In the website you can read more about APPRIS.

APPRIS is involved in the GENCODE project (http://www.gencodegenes.org/) which produces high quality reference gene annotation and experimental validation for human and mouse genomes. So, you will see some differences in the number of genes. GENCODE and Ensembl are collaborating together and there is a merge of gene but Ensembl has more genes obtained automatically.

Are Ensembl IDs stable across releases (in the sense that a particular transcript ID always refers to the same set of exons)?

Maybe this question has to answer someone of Ensembl team but in theory the Ensembl transcripts IDs are unique.

Best Regards,
J

Answer 3

[crisime]

Hi Alex,

Are Ensembl IDs stable across releases (in the sense that a particular transcript ID always refers to the same set of exons)?

Have a look at the transcript examples I posted in this question: Stability of Ensembl and refseq stable IDs

These are examples, where an ENST change quite a lot in different versions on exon-/DNA- and even protein-level, which would lead to totally different variant annotation results for the different ENST-versions.

Most of the time it should be the same exons, but the answer to "... always ... the same ..." must be no.

However, these are not examples, which affect the current principal isoforms. I don't know, whether there exist any. But if you want guarateed exon/DNA/Protein-identity between versions, these examples imply, you have to check for yourself for whatever you need.

Best regards

Chris