Liftover And Sequence Conservation

Question

[epigene]

I'm using liftOver to lift some mouse sequence to human sequence. It seems (I believe so) liftOver parameters will influence the result. So I'm wondering if I can infer conservation between the mouse sequence that are liftover'ed to human and the output human sequence? In another word, can I say if a mouse sequence can be liftover'd to human, these two sequences are conserved? Is there a way to quantify the conservation? I'm feeling if the phastCons score in the mouse sequence is high, that can at least say that the mouse sequence is conserved, but not sure if it's conserved with the human liftOver sequence or not.

I'm interested in testing mouse sequence functions using human cell line. So it's important to know whether I can infer mouse sequence function from doing this kinda experiment in human cell line.

Thanks!

Answer 1

[Istvan Albert]

As far as I know liftover is a tool to convert between different genome builds of the same species and not across species.

To infer conservation look at the results of whole genome alignments.

[Johanna Schott]

I use liftover between species (mouse and human), and as long as the sequence is sufficiently conserved it works very well.

[epigene]

Yeah, liftOver does work between species but I think you need to take it with a grain of salt.. that's why I want to double check the results I'm getting from liftOver. I asked another question to compare two sequence conservation...

[Bert Overduin]

That is what I thought too (so I upvoted your reply), but then it says on the UCSC website: "Batch Coordinate Conversion (liftOver) - converts genome coordinates and genome annotation files between assemblies. The current version supports both forward and reverse conversions, as well as conversions BETWEEN SELECTED SPECIES." ....

[Istvan Albert]

Interesting I did not know that. I always considered it a build conversion tool but indeed even the website has drop-down menus across species.

Answer 2

[Johanna Schott]

As far as I know that PhastCons score is reflecting the conservation in the multiple species alignment and does not necessarily reflect conservation between mouse and human, because many other species are also part of the alignment. But you can have a look at the pair-wise human-mouse alignment of your region in the mouse genome. You go to your region in the mouse genome browser. Within the Euarchontoglire Genomes, Chain and Net Alignments track, you pick the Human Alignment Net track (by clicking on it in the browser), click "View alignment details of parts of net within browser window.", and then you should see the pair-wise mouse-human alignment.

There are alternatives, if you want to figure out whether your sequence of interest is also present in humans and how similar it is to the mouse sequence. It depends on what kind of sequence it is. If it is transcribed and codes for a protein, I would blast the mouse protein sequence to find homologs in the human proteome. For a non-coding RNA, you could blast the RNA sequence against human ESTs. For a sequence that is not transcribed, you can blast against the human genome. These are more open approaches than to look only at the corresponding position of your mouse sequence in the human-mouse genome alignment.

[epigene]

Right, PhastCons scores come from multiple species alignment..

With the pair-wise mouse-human alignment, will the two aligned sequence will always be the same as the liftOver result?

The sequence of interest are random sequences in the genome, mostly non-coding and I don't know if they are non-coding RNA either, probably not. My approach now is to take liftOver result, get the sequence and align the two and see the alignment and similarity. Not sure if this is the best way though.

[epigene]

I just checked the Human Chain/Net track and it's pretty informative. Thanks!

[epigene]

I have a question about the interpretation of the browser. Can you open the link :http://genome.ucsc.edu/trash/hgt/hgt_genome_4fc9_3dfaa0.png

This is showing the region in mouse and its human chain/net tracks. The liftover result matches the mouse sequence to chr15, so it's the second one shown in the chain track and level 2 in net track. I wonder if liftOver results is the second one, why are these one above it in both chain and net track? The first one has double line representation and that would suggest it has a complex structure, right? So I should not rely on the chain/net track solely to pick the best match in human for a mouse sequence?