A Question About Oxford Nanopore and the Cost Model

This is the first mention of how much the much-vaunted Oxford Nanopore sequencing technology might cost (boldface mine):

Oxford Nanopore said last month that it plans to make the MinIon sequencer commercially available through the MinIon Access Program. Customers need to join the MAP, which the firm describes as a “developer-style access program,” and pay a $1,000 fee, which includes the MinIon MkI device and a starter kit with three flow cells, two reagent kits, software, and “ongoing intermittent free supplies.”

…Users can purchase additional flow cells in sets of 1, 12, 24, or 48, at a cost ranging from $500 per flow cell to $900 per flow cell, depending on the number of flow cells purchased at a time.

Oxford Nanopore does not specify the maximum yield of each flow cell, which has been increasing with newer versions. Run time is not fixed, so output varies depending on the length of the run. Last year, the company said its internal record was 1 gigabase per flow cell, and some users achieved more than 500 megabases per flow cell.

Last month, the firm said it plans to introduce a fast mode option, which will increase the data output by tenfold, to up to 40 gigabases for the MinIon MkI.

If we limit ourselves to best case scenario of what is actually in hand, 1Gb of sequence at $500, that doesn’t seem so hot. In the microbial world (cuz humans are boring), that might get you two or three bacterial genomes per flow cell, and that would then require some sort of barcoding method (MOAR MONIES!) to tell samples apart. Otherwise, you are, in the best case, spending $500 on sequencing reagents alone per bacterial genome. You might as well do PacBio sequencing.

Yes, I realize Nanopore doesn’t require a very expensive machine. But if we’re serious about getting this into the clinic, as opposed to doing gee whiz stuff, this needs to be cost-effective (and generate reliable data* of course) at a scale of dozens or hundreds of samples per day. Doesn’t seem like it’s there yet, though hopefully the “fast mode option” isn’t vaporware, which would help lower costs, although require additional molecular biology expenses to barcode DNA from different strains.

There endeth the grumpiness.

*I’m not convinced at all that for routine surveillance (not “whaddya think about these six isolates”) the error rate for Nanopore allows you to identify incipient outbreaks. Too many errors as best as I can tell.

This entry was posted in Genomics, Public Health. Bookmark the permalink.

4 Responses to A Question About Oxford Nanopore and the Cost Model

  1. Pingback: Oxford Nanopore Fans are Asking Three Questions Today « Homolog.us – Bioinformatics

  2. homolog.us says:

    Well, you did not mention their ‘no criticism’ policy but other than that I agree.


  3. Andreas Klostermann says:

    Your information about both the pricing models and yield is out of date. Clive G. Brown said there will be a model where flow cells are kept handicapped, so that you can buy a flowcell for initially three hours of sequencing, very cheaply, and if you need more, you can unlock the cell by paying more. This is meant for situations where there is no need for a couple of gigabases during the lifetime of a flow cell.

    Anyway, the 1 Gigabase yield is on the low side of MinION capabilities. Several MAPers are already getting 2-3 gB. Fast Mode, Chemistry improvements, and the new ASIC with 3000 instead of 512 channels may get you between 20 and 40 gB per flowcell in two days, by the end of the year. Clive Brown says, Promethion may deliver several Terabases in the same time, but I figure this is a lot harder to manage.

    Another important part for clinical sequencing is that it’s absolutely not necessary to assemble a full-fledged genome for identifying a bacterial strain. Depending on applications, pretty much any good longish read will be conclusive.

    So far, I haven’t heard about barcoding to tell apart samples. Most people seem to be content to just run one sample for as long as necessary to get enough data – and because sequence data arrives in realtime this can be determined precisely – and then wash the cell, and start the new sample.

    Depending on improvements, it seems achievable to identify strains in minutes of runtime. Other methods can’t do that currently. Even at the current state, the MinION technology is the fastest way for diagnosis by sequencing, because sample prep is faster, data is realtime and reads are longer.

  4. Bob Chadwick says:

    You are ignoring the fact that amazingly long read lengths are being obtained (100 to 200 kB). This is revolutionary for genetics and the DNA sequencing field.
    Also, the amount of data generated by the MinION is growing exponentially as is the sequence quality.

Comments are closed.