# Can exons be located outside of the coding sequence?

I have a gff file like this (I edited the name):

scaffold_x  source  exon    2987526 2987805 .   -   .   name "foobar";transcriptId 68892
scaffold_x  source  CDS     2987526 2987805 .   -   1   name "foobar";proteinId 68892;exonNumber 5
scaffold_x  source  exon    2988610 2988745 .   -   .   name "foobar";transcriptId 68892
scaffold_x  source  CDS     2988610 2988745 .   -   2   name "foobar";proteinId 68892;exonNumber 4
scaffold_x  source  exon    2989157 2989645 .   -   .   name "foobar";transcriptId 68892
scaffold_x  source  CDS     2989157 2989645 .   -   2   name "foobar";proteinId 68892;exonNumber 3
scaffold_x  source  exon    2990903 2991541 .   -   .   name "foobar";transcriptId 68892
scaffold_x  source  CDS     2990903 2991541 .   -   2   name "foobar";proteinId 68892;exonNumber 2
scaffold_x  source  stop_codon  2990903 2990905 .   -   0   name "foobar"
scaffold_x  source  exon    2991587 2992430 .   -   .   name "foobar";transcriptId 68892
scaffold_x  source  CDS     2991587 2992430 .   -   0   name "foobar";proteinId 68892;exonNumber 1
scaffold_x  source  start_codon 2992428 2992430 .   -   0   name "foobar"


So the start_codon is 2992428 2992430 and the stop_codon is 2990903 2990905; I would therefore expect all the exons within this range, however, e.g. exon 5 is found in 2987526 2987805 which is clearly outside this range.

Is there something wrong with the gff file itself or is there a plausible explanation for that?

EDIT:

In the corresponding protein fasta I do find an amino acid sequence associated with this prtein ID and name:

>68892|foobar
MHTGD...

• Thanks for editing, but that doesn't tell us anything. We need to know the accession of the protein and/or gene you are looking at. What is 68892? Is that the gene ID? Also, why would there not be a protein sequence associated with it? It is only the UTRs that are not translated, the rest of the exons will be. Oct 20, 2017 at 8:59
• @terdon See my comment below your answer. And thanks for the hint; there are indeed exons in the "correct" region which can then be translated to a protein. Feel free to expand your answer and then I am happy to accept it.
– Cleb
Oct 20, 2017 at 9:03
• I don't get what is confusing you, to be honest. I'd be happy to expand, but I don't know in what direction. It might be simpler if you come into chat for a second and ping me (@terdon) there so we can figure out what is confusing you. Oct 20, 2017 at 9:04

Yes, of course. Exons are not limited to the protein coding regions. Many UTRs are in exons. In fact, you even have various cases of UTRs being multiple exons, and being spliced.

What is strange in your file is not so much that you have exons beyond the stop codon, but that you also have them marked as CDS (coding sequence). That isn't possible, no. While there can indeed be exons in the UTRs, those are not coding and shouldn't be called CDS. That's probably just a minor nomenclature issue though and what you are looking at is a spliced UTR.

So, the exons (or the parts of the exons) that fall between the start and stop codons will be translated, while those that fall outside that regions will not be (UTRs).

For example, this is the exonic structure shown for ENST00000617185, one of the transcripts of the human P53 gene:

The boxes are exons and the lines are introns. The colored boxes are protein coding exons, while the empty ones are UTR exons. Note how they are also spliced and that this particular transcript has 3 non-coding exons on the 5' end and one on the 3' end.

• @Cleb no, UTRs are not translated. But I can't really help if you don't give us the gene you are looking at. Please edit your question and tell us where the data is coming from so we can also compare your gff to the predicted protein. Oct 20, 2017 at 8:54
• Thanks already for your answer (+ 1). So if I understand correctly, these regions should not encode for a protein, correct!? Funny thing is, however, if I look at the corresponding protein fasta, there is an amino acid sequence associated with it (see my edit). But, as it is not a coding sequence, this should not b the case then?
– Cleb
Oct 20, 2017 at 8:55
• Unfortunately, it is not publicly available, when I google for the gene name I don't get any hits. But for now it is already good to know that there is something fishy(?) going on so that I can contact the authors; just wanted to make sure that I don't miss something obvious.
– Cleb
Oct 20, 2017 at 8:59
• @Cleb there is nothing fishy going on at all. All (or, at least, the vast majority) of protein coding transcripts have untranslated regions (UTRs) and this is what you are showing. That doesn't mean that all of the transcript is not translated, only that the UTR regions are not translated. And what do you mean it isn't publicly available? Is this a gene prediction? All genes are publicly available. Oct 20, 2017 at 9:01
• Yes, thanks for clarifying, I am not very familiar with this topic, so I probably ask a couple of silly things :) Well, I got the file from a website where one can only download files if one is registered and as far as I understood one is not allowed to distribute the information without permission.
– Cleb
Oct 20, 2017 at 9:07

Those are the untranslated regions (UTRs). All mRNAs have a 5' UTR and a 3' UTR. These give the ribosome something to grab onto and often contain important regulatory sites such as miRNA target sites.

• I've often wondered idly about that. Are you 100% sure that all mRNAs will have two UTRs? Or, at least, all mammalian mRNAs, to keep things simple. The vast majority do, of course, but this is biology, the land of shades of gray and exceptions. Are there absolutely no mRNAs without UTRs? Oct 20, 2017 at 8:06
• Not all have them identified (yet) but functionally I think that UTRs are necessary. I'm no expert on translation, but even ignoring the regulatory aspect, I think they're necessary just to allow the ribosome to attach on. Oct 20, 2017 at 8:28
• Yes, that's what I'm wondering about too. I'm not a translation expert either, but I don't remember any such function. That, however, in no way implies that there isn't one, which is why I thought I'd ask. I'm pretty sure I've seen transcripts with no annotated UTRs, but as you say, that might just be a fault in the annotation rather than an actual biological phenomenon. Oct 20, 2017 at 8:39
• Thanks already for your answer (+ 1). So if I understand correctly, these regions should not encode for a protein as it is a untranslated region, correct!? However, in the corresponding protein fasta file, I find an amino sequence associated with the name and protein ID (see my edit). How can this be?
– Cleb
Oct 20, 2017 at 8:49
• Yes, the coding region encodes a protein. At either end of the coding region is a UTR. Oct 20, 2017 at 9:03

Although the explanation that UTR (untranscribed) regions can consist of multiple exons covers most situations, I think it is good to mention stop codon readthrough.

This would result in a CDS (coding sequence) part of your mRNA (messenger RNA) after a stop codon.

However, following the gff3 format specification, this should probably be encoded differently, using SO (sequence ontology) terms like SO:0000697 (gene with stop codon read through) and SO:0000883 (stop codon readthrough).