Now, he’s only shown us a single read, a roughly 8kb read from Pseudomonas aeruginosa, so there’s only so much that can be gleaned from it. The bloggers over at Next Gen Seek did a nice job analyzing what they could from a single read (with a BLAST result that took 30 min!). The main takeaways are:
- The data are here and they are real!
- The read length is impressive (although not beating PacBio yet)
- The error rate seems pretty high, with lots of deletions and mismatches.
But there are still many unanswered questions:
- How typical was this one read (and out of how many)?
- What is the range of read lengths?
- What is the overall error rate?
- Is the error systematic or stochastic?
This last question is probably the most important one. Despite its high raw error rate, PacBio has a very nice stochastic error profile that ends up producing very low error rate consensus sequence. Early presentations from Oxford Nanopore have implied that the MinION error profile is more systematic, meaning that oversampling won’t solve the problem. They’ve hinted at some methods for getting around this issue, so we’ll have to wait and see.