Ensembl protein counts

Question

[cdsouthan]

Is it possible that some proteins annotated in Gene counts (Alternative sequence) are different from the primary sequence set ?

Answer 1

[Emily]

Yes. Patches may contain novel sequence that allows new genes to be annotated, or fixes in sequence which will give different protein sequences. Haplotypes may contain genes that are not found on the primary assembly, and again different sequences for the "same" gene.

[cdsouthan]

Thanks, next Q please (same topic) I noticed recent protein numbers are identical across releases

Ensembl 85: Jul 2016 19915 Ensembl 86: Oct 2016 19915 Ensembl 87: Dec 2016 19915

Does this mean there was no protein re-build in those releases? or they just fell out as the same? ie all scaffold changes were outside exons?

[cdsouthan]

Next Q please, I'm puzzled by the 526 read-through human proteins since no other pipeline specifies these. Any chance of the FASTA sequences? (have asked Helpdesk so if this is in hand with them that's fine)

[Emily]

Ben will get back to you when he's sorted it out. Turns out how these are identified is not straightforward.

[cdsouthan]

Seeing as I may finally get round to revising https://www.ncbi.nlm.nih.gov/pubmed/28529709 I would appreciate if Ben or anyone else could send me sequences for what are now 562 read through human proteins

[Emily]

There was no rebuild. If you look at the dates on all the pages, you'll see that the genebuilds were all completed June 2016.

[cdsouthan]

OK got it, so that confirms some releases have assembly changes but do not rebuild the protein set

[cdsouthan]

OK so the protein rebuild dates on the FTP sites are:

12/09/2014

06/05/2015

12/6/2016

And therefore the small protein number differences I see in the archive between these dates are just patch changes?

67 May 2012 22369

74 Dec 2013 20805

75 Feb 2014 20342

76 Aug 2014 19889

77 Oct 2014 19855

80 May 2015 19781

81 Jul 2015 19783

83 Dec 2015 19801

87 Dec 2016 19915