Ion torrent for insertion/deletion

[arrayprofile]

Hi, I have a question: because Ion torrent doesn't do paired end sequencing, does this mean the technology cannot be used for insertion/deletion detection?

Thanks

John

[krobison]

What size insertions & deletions are you looking for?

For 1-2nt indels, IT is very difficult due to the homopolymer counting issue

For longer small indels, IT can work -- except there will be noise from the homopolymer counting issue. In some recent data, we definitely spotted a known 5nt deletion -- except in some reads it is called as a 4nt deletion due to the counting issue.

Paired ends are not magical indel detectors, but do enhance your sensitivity for indels of certain sizes. Reads from any platform can detect an indel by crossing the indel boundary; paired ends (and mate pairs or strobes) simply enhance sensitivity by scanning a larger region.

It is surprising that ABI hasn't retrofitted all of their SOLiD mate pair protocols to the Ion. Perhaps they need longer reads to make it work.

[arrayprofile]

Thank you K! If I understand correctly, you don't really need paired end to detect indels if there is no homopolymer issue. But IT had serious issue with small indels because of homopolymer.

For Illumina platform, am I correct that paired end reads only help indel detection for large indels because the fragment size is only approximate (~300 bp), a few nt indel won't be identified by checking the size between the 2 paired reads?

Thanks

[epistatic]

Mate-pair libraries are a bit painful and expensive to produce. With the template on the bead, Ion would need to have a way to melt off the strand and reprime, or do a mate-pair bidirectional read. I'm sure this is an application being developed, the Paired-End ability of the MiSeq is seen to be a benefit over the PGM.

[nickloman]

As other commenters have stated, looking for 1/2 base indels is not the easiest task on Ion Torrent because of homopolymer counting issues which will produce false positives. If you restrict your hunt to regions away from homopolymers you may find the false positive rate goes down.