Coding sequences of genes have a certain degree of evolutionary conservation, so that comparisons based on sequences (BLAST, HMMER etc) can be informative. Generally speaking, the more two species are phylogenetically close to each other, the more their genes will be similar to each other. To add to that, different genes will have different degrees of sequence conservation - for instance, histons are very conserved and so on.

Do these considerations hold true for untranslated regions of genes (both 5' and 3')? To what extent can an alignment of UTRs be meaningful? Are there any very conserved UTRs as there are very conserved CDS? I know that there are some conserved motifs in 5' and 3'UTRs, but what about the overall sequence?

  • 2
    $\begingroup$ Depends on the gene but, in general, less than CDS $\endgroup$
    Commented Jul 3, 2019 at 17:21
  • $\begingroup$ Can you please clarify what you meant by "for instance, introns are very conserved"? Did you mean exons? Introns are generally less well conserved than exons. Also, introns are parts of genes not genes so this doesn't seem to be a relevant example to give. $\endgroup$
    – tyersome
    Commented Jul 3, 2019 at 22:08
  • $\begingroup$ @tyersome Oh sorry, I meant "histons", not "introns". I just wanted to make an example of a very conserved gene. $\endgroup$ Commented Jul 4, 2019 at 11:08
  • 2
    $\begingroup$ I assume you mean histones not histons. In which case why don't you download sequences of human and mouse histone mRNAs from Genbank and compare them by one of the many online programs available to do so? The overall conservation is likely to be low (it is for actins), but this does not preclude more subtle conserved features. $\endgroup$
    – David
    Commented Jul 4, 2019 at 20:42
  • $\begingroup$ @David ...yes, sorry, English is not my first language so I get confused. I'm not interested in levels of conservation in histone sequences, I'd like to know how conserved are UTRs at a broader phylogenetical level. I found this article (bmcgenomics.biomedcentral.com/articles/10.1186/…) about 3'UTRs in cichlids and non-cichlids, but not much more. $\endgroup$ Commented Jul 5, 2019 at 14:14

1 Answer 1


This is a question specific to the organism, assembly, and genes of interest to you. You can answer this yourself with a bit of elbow grease.

You could measure mean phyloP or phastCons conservation signal over UTR-annotated intervals, and compare it against the background mean conservation signal over random, mappable regions of the genome of similar length (i.e., calculate the mean and variance in UTR length and sample mappable regions of the genome for intervals with the same distribution).

At a very basic level, you could perhaps apply either the U-test or Wilcoxin rank-sum test, depending on your assumptions of dependence, whether your background does or does not include UTRs.

One source of phyloP and phastCons signals for various genomes and multi-genome alignments is UCSC Goldenpath. For example, here are phyloP 100-way alignments to hg38. You might get UTR annotations from Gencode, assuming you are working with human or mouse. To do interval overlaps and mapping/statistics operations, you might use gff2bed, bedops, and bedmap in the BEDOPS toolkit(*), and then use R to do statistical tests.

Disclaimer: I am one of the authors of the BEDOPS toolkit.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .