Using 50bp short-reads to find a translocation

[agc]

I know of a translocation that occurred (I also know the sequence that was translocated) in my sequenced DNA, but I'm not sure where it was translocated to, and the reads of the translocated sequence line up against the reference in the original location. Is there any way to find a translocation using short-reads?

[poisson200]

Hi Agc,
It would help people if you supplied more information;

1) Is it genome re-sequencing or transcriptome data
2) Is it paired end or single reads
3) What is the coverage/number of reads generated
4) What species is it

I don't know if there is a ready made solution for you but my bodge up and leg-it approach would be:

1) Reads that span the translocation won't map to your reference sequence. E.g. Maybe as, for some reads you have, say the first 25bp could map to chromosome 1 and second 25bp to chromosome Y, using human as an example.
2) Assuming my example is similar to your situation, I would take all reads that did not map; for each read here take the first 21bp and the last 21bp of the read and map to genome (maybe with BLAT or BLAST, altering paramters). See which first 21bp and last 21bp map to different chromosomes, there will be your candidate translocation regions but hopefully one stands out as being real (lots of reads mapping to them).
3) Take the sequences of putative translocation, 49bp from each chromosome and make a pseudo translocation sequence BLAST database
4) BLAST all reads against translocation sequences and record those reads that align over full length of the read. The most likely real translocation will have most reads mapping which at least overlap the translocation point by 1 base

This is my rough approach, which will get you the right answer but involves a bit of BLAST, BLAT, Perl/Python magic and some result filtering.

I predict some better experts of NGS know of better/easier solutions, probably with some already developed software. So give it a day or two before embarking my solution.

Good luck.

:-)

ps. If it is paired end, this will help a lot as one mate pair will map to one chromosome and the other mate pair another chromosome (there should be definitely software to help do that) or just parse the SAM output from TopHat or BOWTIE.

[agc]

Thanks for the quick reply!

1) Genome re-sequencing
2) Single ~50bp reads
3) Not sure where I can obtain that information.
4) S. Cerevisiae

The translocation occurred within the same chromosome, but I'll try to develop the idea of using the unmapped reads. Although I'd find some sort of ready made solution / any other suggestions very helpful.

[poisson200]

It will still work by mapping the 21bp read ends and find those that both ends map to this chromosome in question. Look at the distance between the two ends. Most reads should map each side of translocation site have similar distance between ends.