How To Liftover Genomic Regions From Dm3 (Drosophila) To Mm9 (Mouse)

Question

[biorepine]

Liftover tool can not do this operation. Does any know how to handle this problem ?

Thanx

[Sean Davis]

Have you asked UCSC if they can provide the files for you? They have been known to respond to such requests.

[biorepine]

I haven't. But I will now. Thanx!

[lh3]

What is your intention? I doubt you can get anything usable. The distance between the two is so large that essentially no synteny blocks exist except a few well known gene clusters; even for the few that exist, they are hardly identifiable from nucleotide alignment.

[biorepine]

You might be very wrong. This paper "A Global In Vivo Drosophila RNAi Screen Identifies NOT3 as a Conserved Regulator of Heart Function", reports thousands of orthologs between mouse and drosophila. But the analysis limited to only genes. In my case I would like to know every genomic region that is a candidate ortholog.

[lh3]

I made treefam and was involved in developing Ensembl-Compara. I of course know there are thousands of orthologs between human/mouse and fly. However, these orthologs can hardly be identified from nucleotide alignment and without long synteny blocks, the UCSC aligner will have very high error rate. Note that I am not saying there are no highly conserved regions, I am saying that the generic UCSC pipeline cannot give you reliable alignment. You need specialized tools to identify conserved regions given such long divergence.

[biorepine]

Sorry if I sound bit odd in the previous comment. I am not the expert in this ortholog study. I will note your points. Thanx!

Answer 1

[KCC]

This isn't a great answer and I would be highly suspect of the results. However, you can do dm3 to dm2, dm2 to hg18 and then hg18 to mm9.

The liftover files can be found here:

http://hgdownload-test.cse.ucsc.edu/goldenPath/dm3/liftOver/

http://hgdownload-test.cse.ucsc.edu/goldenPath/dm2/vsHg18/

http://hgdownload-test.cse.ucsc.edu/goldenPath/hg18/vsMm9/

Here I am assuming regionsdm3.bed are your fly genes that you want lifted and regionsmm9.bed is your output. This should get you started, if you just want to get some quick answers.

liftOver regionsdm3.bed dm3ToDm2.over.chain.gz regionsdm2.bed failed1.bed

liftOver regionsdm2.bed dm2.hg18.all.chain.gz regionshg18.bed failed2.bed

liftOver regionshg18.bed hg18.mm9.all.chain.gz regionsmm9bed failed3.bed

However, lifting between species is already dodgy, as the liftOver tool will tell you. So, lifting through another species as an intermediary seems double dodgy to me, but sometimes something is better than nothing.

I support contacting UCSC directly as a better solution.

Answer 2

[Hugo A F Santos]

Dear all,

Sorry to revive such an old post, but I am still puzzled on the existence of a more defined answer, as of today, for this issue.

Is it possible - and by what means - to accomplish what the original author of this thread asked? I am in the situation of wanting to compare differential exon usage/ splice isoform prevalence in terms of consistency between the human species and Drosophila - but, as you have pointed out, the entire structure of the gene (size, number of exons and so on) is completely different; not to mention that it is quite hard to link one particular exon in Drosophila to one in Human even upon "smaller" genes with < 5 exons.

Thank you for your time.

Best regards,
Hugo