Monday, 19 January 2015

Illumina's new sequencers - what will we do now?

I was as surprised as everyone else with the most recent announcements by Illumina (JP Morgan post), especially their rapidity given that V4 chemistry is just 6months old; and I quickly sketched out my initial thoughts at the beginning of the week. There's been lots of other coverage, posts (MassGOmics O, Mick, LabSpaces), and activity on Twitter.

I've now had a chance to digest my thoughts and plan out what I hope we can do here at the CRUK Cambridge Institute - "I'll take two please."

Why buy 2: The HiSeq 4000 is a big step forwards, costs are likely to drop again making the 2500 less attractive (and sending our 2000 to the skip - For Sale anyone) and output goes up decreasing the number of lanes we need to run. The patterned flowcells also mean we can forget about being quite so careful with library quant. All of this is good news but it does mean we've got to finance new instruments.

Ideally we'll replace both 2500's in my lab with 4000s and move any rapid work over to our NextSeq once V2 chemistry is available. MiSeq would be reserved only for users looking for the cheapest run, but I suspect most people will be happy to migrate to NextSeq. So potentially we've got rid of three sequencers (1x2000, 2x MiSeq), increased capacity, lowered costs and freed up space in the lab. Now all I need to do is convince my boss!

I want two because when a single machine breaks down we have nothing. The reliability of HiSeq is not something I've posted about before, and believe me there's been times I've wanted to shout from the rooftops! But the move to 2500 and V4 has been a revelation. The 2500 hardly ever breaks down and V4 chemistry works almost every time. I suspect a lot of this has to do with the reduction in run time from 12 to 6 days, runs simply have less time to fail in. I should take the opportunity to say "well done" Illumina and thanks to the support team for your efforts.

Patterned flowcell: The big excitement from my perspective is the advent of patterned flowcells for the rest of us. These are an incredible piece of development, we can reduce the yield variability between runs, and almost forget about library quantification by qPCR (Kapa revenues might drop). However we will need to test many of our protocols as only a few are supported by Illumina. The 4000 also means that genomes might well be all you can do on X Ten for the next 1-3 years.

I'm hoping we'll be able to try some cool experiments on the patterned flowcells. Insert size is constrained due to "pad hopping" and read-duplication - but I am sure there is room for invention and we are not constrained by Illumina's desire to have Q40 sub 1% duplication on every run. Some applications might be better done with what could be considered poor quality data.

Illumina's patterned flowcells

What about Broad and Sanger: I suspect the really big impacts will be felt in the larger labs - certainly Genome Centres but also possibly labs with three or four+ 2500s. The Broad currently has around 14 X Tens and many more HiSeq's. HiSeq 4000 with patterned flowcells allows all applications to be run on one technology; Human genomes on X and everything else on 4000. This probably means the number of 4000s required is very low indeed, even at the Broad! Big savings in sequencing consumables, service contracts, space, and staff. Moving to HiSeq 4000 for some of the larger labs is likely to be not just cost-effective but possibly cost-neutral or even cost-negative.

The impact on the market: Ilumina's stock price has dropped about 10% since JP Morgan, seems crazy at the outset after such big announcements. However the disclosure of reagent pull through (Illumina's longer term revenue) is surprisingly low. And marketeers need to figure out what some of these announcements mean for the people using the technology. If we all switch then great, but many of us might switch off because we can't afford to get new instruments into our labs.

1 comment:

  1. Geneticist from the East20 January 2015 at 08:13

    I think Illumina should have a program to let 2000/2500 owners to sell their machines back to Illumina for a price. Then Illumina can certify them and re-sell them at discount to others. This helps small labs to get 2000/2500 and big labs to have funds for 4000.