How to detect a replacement gene/region

[westerman]

Illumina PE sequences.

I am having trouble finding a program that detects a replacement gene/region. Either I need a different program than the ones have been running (Pindel, Breakdancer, indelMiner) or better command line parameters. The programs detect deletions and short insertions just fine but don't tell me "this gene has been deleted and replaced with a different gene from outside the organism".

It is easy to see the region in IGV or other visualization program. Visually the region looks like a deletion however unlike a normal deletion where it is possible to see the two PE reads map to either side of the break and see a single read get broke, in the replacement case one side of the PE reads maps while the side does not.

Any suggestion as to a program to use?

Thanks,

[westerman]

As an example of what I mean, here are two IGV screenshots.

The first ("deletion_found") show the reads mapped to the genome. You can see paired reads spanning the deletion (the thin lines between the reads). There are also some reads that are split apart. Not that coverage goes down on either side of the deletion breakpoint.

The second screenshot shows a similarly sized region (~4500 bases) and has a similar coverage pattern. However note that there are no pairs spanning and that broken reads then to have mismatched bases near the breakpoint. The lack of paired reads is because the pair is part of the replacement gene and since said gene is not part of the reference then no mapping occurs.

Pindel, for one, picks up the first region (deletion) just fine but skips over the second region (replacement)

I am using Samtools to pull out places where there are a bunch of mapped but mate unmapped reads. This might work for my purposes. I was just hoping that there would be some program that can pick up replacement genes for me.

[westerman]

In the end I used a combination of cortex as well as pindel for determining indel/replacements. Neither program seems to be perfect and I am not totally satisified with the results in that I have the feeling that I may have missed something. Annotation was done with snpEff.

BTW: I really like the cortex manual. Well done.