How to align long primary transcript sequences?

Question

[patricia_icbc]

Hello,

Does anyone know how to do pairwise alignment of very long primary transcript(with long introns, e.g. 6k) sequences?

Will lastz/chain pipeline work? If so, could anyone please tell me how?

I have been bothered by this question for a long time and really appreciate any of your help.

Tricia

Answer 1

[thackl]

If you want to align unspliced transcript vs. unspliced transcript, try exonerate (exonerate --model genome2genome0. It models introns on both, query and target sequence.

[patricia_icbc]

Thanks, thackl.

I tried to use exonerate with --model genome2genome, but the output is just alignment of some fragments. Can exonerate generate an alignment with two sequences aligned from end to end?

[thackl]

I was actually hoping that genome2genome would do this But you are right, as a second look at the available models in the manual confirmed - genome2genome is not a global alignment mode.

Exonerate can run global alignments, e.g. --model affine:global, however, this mode does not model introns...

There might be a few ways to improve your result (not sure though, if this gets you far enough...)

1) use genome2genome, but try to prevent exonerate from splitting the alignment. Have a look at --refine, --exhaustive and Intron Modeling Options

2) use affine:global but maybe with adjusted --gapextend costs to allow for long gaps (and also use--exhaustive and --refine)

[thackl]

It would also help to know a little bit more about your sequences. Are you aligning transcripts from different organisms, or different paralogs, or ..? And are you expecting minor differences, primarily in introns or in exons, or are you looking for structural differences?

Answer 2

[Prakki Rama]

You can also try mapping using GMAP if you align to genome.

[patricia_icbc]

Thanks, Prakki.

Cound GMAP align genomic sequences?

[Prakki Rama]

It seems it aligns only cDNA and ests. Check the link I provided for more details.

[patricia_icbc]

Then it may not solve my problem. Thanks all the same.