How to visualise RNAseq baits in a genomic context

Question

[Duarte Molha]

Hi everyone

I am in the process of creating a RNAseq bait design and I have created baits that span splice junctions to better target a specific transcript of interest. I was hoping to find a way to visualise these reads in a genomic context on a genome browser like IGV to better assess how the bait coverage looks on the genes on interest.

Anyone know how to best achieve this?

Many thanks

Duarte

[Nicolas Rosewick]

make a bed file with the bait coordinates and open it in IGV

[Duarte Molha]

Unfortunately.. that is not as easy as I would like...

I start with the complete coding sequence of the transcript I am interested in, I then tile across that sequence to create by baits. This means that each bait will have, potentially overlapped 1 or more splice junctions.

The process of then going from that tiled bait sequence to the exact genomic coordinates it overlaps is time consuming.

I was hoping there would exist some software that I could use where I could provide the genomic coordinates of the transcript and its genomic sequence together with the bait sequence and it would give me the genomic coordinates that the bait spans.

This I could then input into IGV

Answer 1

[h.mon]

Map your bait sequence with a splice-aware mapper (BWA, GSNAP, GMAP, STAR, etc) and open the alignment on IGV.

edit: BWA is able to perform split alignment, but it is not splice-aware. And GMAP is better than GSNAP for the task at hand, and its default intron length is large enough (1,000,000) it shoud work for you.

[Duarte Molha]

Thanks... I just tried bwa mem ... but it simply soft-clipped the parts of the bait that align after the slice junction. How to you force it to accept mapping large mapping distances so that it ignores the insertion if the intronic region in between the 2 segments of the bait?

[Duarte Molha]

Many thanks... I will try GMAP

[Duarte Molha]

I just used GMAP and it worked like a charm.. however @h.mon I do get a few baits that where the splice junction is very close to one of the extremities where the aligner will align that small chunk to the intronic region instead of the correct region on the other slice adjacent exon (see screenshot).

Since these are baits and not reads, they match perfectly to the reference sequence, do you know of a way to force GMAP to look for a match elsewhere?

http://awesomescreenshot.com/0884zq2590

[h.mon]

Sorry I missed your comment. I don't know if it will work, but you could play with the -x parameter:

  -x, --chimera-margin=INT      Amount of unaligned sequence that
                                triggers search for the remaining
                                sequence (default 30). Enables
                                alignment of chimeric reads, and
                                may help with some non-chimeric
                                reads.  To turn off, set to zero.

Another idea would be to try another program, such as Blat. But if only a few reads have this problem, it is probably easier to fix then by hand.