yesterday John, a postdoc of Mark's, gave a cool journal club on this paper:
Koren, Schatz, Walens: Hybrid error correction and de novo assembly of single-molecule sequencing reads. doi:10.1038/nbt.2280
http://www.nature.com/nbt/journal/vaop/ncurrent/full/nbt.2280.html
They try to improve PacBio long DNA sequencing reads which have a massive error rare of ~15% with short Illumina (and possibly 454) reads. And yes, that seems to work. However, there are some crucial data missing from their analysis. Then some parts of the paper are a strange accumulation of we did this, that, and that, and yes this as well - and nobody quite gets why. Well and in the very end, it's not really cheap and also not very fast. Although PacBio is fast, one has to generate the Illumina data anyway, which cost time. Then the PacBio throughput is still very low and consequently to do this for a reasonable sized eucaryotic genome should easily get you into regions above EUR 30K or so. All in all, it's pretty much a good commercial trying to rescue PacBio.
No comments:
Post a Comment