Friday, 17 February 2012

Oxford Nanopore did not disappoint

AGBT is over as far as many attendees are concerned, the sequencing landscape looks like it might be about to ramp up to the next gear. Again!

The new MinION (read about the USB seqeuncer later on) will cost $500. assuming everyone at AGBT buys one ONT get $5M in rvenue form one conference. This is before anyone does anything real with the technology.

At least one lucky punter got a sneak preview. Nick Loman, my collaborator on the Google Map of seqeunceres, got an early interview with Clive, Zoe and Dan at ONT. Read his post here.

Clive Brown from ONT presented “Single Molecule ‘Strand’ Sequencing Using Protein Nanopores and Scalable Electronic Devices” at AGBT in a 20 minute slot. It was 20 minutes with lots of information, much of it jaw dropping if it delivers as well as everyone hopes. The company also have a press release which says “[Clive] outlined the Company's pathway to a commercial product with highly disruptive features including ultra long read lengths, high throughput on electronic systems and real-time sequencing results. Oxford Nanopore intends to commercialise GridION and MinION directly to customers within 2012".

I can't resist adding this image of Clive... I lifted it from Bloomberg (sorry Bloomberg). He looks like the cat that killed the rabbit! And yes, that is a DNA sequencer he is holding in his fingers.

ONT have commercialised DNA 'strand sequencing', not the exonuclease method Illumina licensed for $18M.

The sequencing chip:
ONT are not using any published chemistry in their system. The chips apparently have an ASIC core with a membrane covering them supporting the nanopores. The aim is to release chips with 4000 or 8000 pores each. The enzymes used in the sequencing run at 20-1000bp/sec. Currently ONT are only using the change in current to differentiate bases, dwell time has been proposed as a way increase discriminatory power and sequence MeC, etc.

The sequencing is done by reading 3bp Kmers, i.e. the pore is large enough to hold three bases inside. These three bases give a characteristic signature and as the strand translocates one base at a time sequential Kmers allow discrimination of the sequence.

Sample prep: The current preferred sample prep is to add a hairpin to the molecule to allow sequencing in both directions. Load the first strand, run through the hairpin and start reading the second strand all in one contiguous read. DNA can be fed through 3’-5’ or 5’-3’.

The chips are very stable and can be loaded with blood to allow sequencing of DNA in solution with no sample prep at all. Clive is going to get a reputation as a mean b*****d after describing sending a vegetarian colleague to an abattoir to collect blood, then getting it from someone’s pet rabbit and finally sending another ONT employee into a river to collect raw sewage. All in the name of progress ;-)

Genome sequencing: Clive discussed a couple of projects. First ONT have sequenced PhiX. But they did it in a single contiguous read. First pass sequencing showed about a 4% error rate but the hairpin double-strand sequencing reduced this to 0.1-2% and the errors are in know locations of the genome as some Kmers are know to be “wobbly”. Next they sequence Lambda, all 42Kb in single reads, and as 5 and 10Kb fragment libraries. The Lambda sequence was done as a 100kb read, 42kb from each direction. The most interesting thin here was that the quality of the 1st, 10000th and 40000th bases is the same. Unlike all other systems that show some form of decay in quality, ONT may be able to give us almost infinite read length, whole chromosomes perhaps?

RNA-seq and Epigenetic modification:
MeC and hMeC have been looked at but many other could potentially be analysed.

Clive briefly mentioned RNA-seq with no cDNA conversion but immediately pointed out thsat though this had been demonstrated there were no plans to commercialise the application at this time. An interesting feature about the ONT methodology is that once one strand has been sequenced a new one can load in and begin sequencing. I can imagine this is going to allow some really cool RNA-seq once they finish the development of that, even with only 4000 pores per chip, at 1000bp/sec you could sequence nearly 3M transcripts of 5Kb average length. With no cDNA intermediary!

GridION: The GridION was presented in a 20-node cluster format that, when loaded with 8,000 pore chips, could sequence a human genome in just 15 minutes. The GridION will be available first with 2000 pore chips, 4000 and 8000 will come in the mid-term. Using the “Run until” technology, a user can specify how much data they want form a samples and leave the instrument to run for just long enough to get all the data. I did not get a chance to see the pricing

The star of the show - MinION:
This is a disposable USB sequencer that will generate up to 1Gb of sequence in the field. Just plug into your laptop and sequence! The chips are capable of 5-25Gbp per day but runs are limited to 6 hours due to chemistry.

The possibility of sequencing in the field brings a whole new dimension to disease and agricultural research. Being able to swap a patient or collect a fungal sample and sequence to identify which drug or fungicide should be used is going to revolutionise research, health and agriculture.

We can't place an order just yet. I certainly hope to start some early access collaborations (hint, hint) and am still not really sure what difference this is going to make to the world.

PS: sorry if there is anything incorrect in this post, there was a lot in Clives talk and I have been grabbing a few minutes here an there to get the post together.


  1. Cheers to you and Nick as always James! All fabulous stuff we can look forward to. Was anything mentioned about the possibility of protein or small molecule sequencing using similar techniques (not that this isn't impressive enough!)

  2. You know that possibility of sequencing in the field brings a whole new dimension to disease and agricultural research. Keep it up!!