I am studying the transcriptome of Arabidopsis. Interestingly, the 5' UTR of the latest annotation is usually too long.
Here is an example (this gene is right to left).
You can see both EST (orange bars) and the RNA-seq reads cover a much smaller region of the annotated 5' UTR. Unfortunately, many 5'UTR of the genes have the same problem. Is there a way to fix that? I was wondering if any assembly tools can help me to get the correct 5'UTR ranges. Thanks!!!
$\begingroup$
$\endgroup$
10
-
$\begingroup$ What evidence do you have that the gene stops earlier? This observation could just be due to selective amplification of the 3' end in the RNASeq reads. $\endgroup$– gringer ♦Commented Feb 26, 2018 at 3:18
-
$\begingroup$ I think it starts late (not stop early) as I mentioned 5'UTR, not 3'UTR. $\endgroup$– l0110Commented Feb 26, 2018 at 3:23
-
$\begingroup$ Do you really think standard RNAseq is perfect enough to judge the start of a transcript with single-base resolution? There's a reason that people have developed methods explicitly for that. $\endgroup$– Devon RyanCommented Feb 26, 2018 at 7:38
-
$\begingroup$ It is not only my data. It is also from many people's data. There is no read at 5'UTR in the old EST data, and in >100 set of RNA seq data. There is just nothing there. I do not understand why those upstream sequences are annotated as 5'UTR, and it could mislead many sequence analysis. $\endgroup$– l0110Commented Feb 26, 2018 at 14:02
-
$\begingroup$ @Devon Ryan, Could you please enlighten me the reasoning (why did they give a long 5'UTR with no evidence?). I am asking this because there are lots of genes behaving like this. I am afraid it will cause a problem in motif finding. Thanks! I updated the image to include the EST data. $\endgroup$– l0110Commented Feb 26, 2018 at 14:18
|
Show 5 more comments
1 Answer
$\begingroup$
$\endgroup$
- Genes, particularly transcribed genes in Eukaryotes, can have alternate start and stop sites (and splicing). Alternate transcriptional start sites may be found in different tissues or at different developmental periods due to alternate transcription factor binding sites being used. The only reliable way to determine the transcriptional start and stop sites is by read coverage.
- Genome annotations are generally the product of bioinformatic pipelines and are mostly predictions based upon computational models. Different gene-finders may call different boundaries; for example, in eukaryotic genomes, an exon may be assigned to the wrong up/down-stream gene. Arabadopsis, as model organism, has had it's gene boundaries considerably refined/corrected/validated, but even so, there could still exist errors in the annotation or, more likely, the boundaries indicated are the most common start/stop sites (see point 1 above).
In short, your UTRs are whatever your experimental data (reads) show it to be. Don't worry too much about it not matching up to the standard annotation boundaries.
