Thursday, 16 January 2014

HiSeq X Ten: only Human Genomes?

Today I thought I'd give my first impressions on what HiSeq X Ten might mean, the dust has certainly not settled and yesterday there was lots of buzz about the new instruments from Illumina (see the bottom of this post for a round up of news). I’ve been reading the HiSeq X Ten datasheet and a couple of things jump out as significant changes: run speed and output per flowcell.

Update: I'd also point readers to this post by Shawn Baker (CSO at AllSeq)  where he points out that Illumina are using a four year lifespan to get the $1000 genomes. We managed to keep our GAIIs for about 4 years, and HiSeq was only 2 when 2500 was announced. Hopefully X Ten can go the distance?

Obviously the biggest issue for mere mortals like myself is the need to find $10,000,000. Three companies were announced by Illumina yesterday, Macrogen, the Broad Institute, and the Garvan Institute of Medical Research. Notably absent form that list were WashU, Baylor, the Sanger and BGI. I think we can devote a whole post to what BGI might be doing if they don't buy into HiSeq X Ten, perhaps a subject better discussed over a bottle of Tsingtao after hours at AGBT. 
The other notable exception from my perspective was GenomicsEngland. How else is the UK, sorry I mean England and NOT Scotland, going to seqeunce 100,000 genomes? Hmm, surprisingly (or not) the HiSeq X Ten makes David Cameron's announcement of £100,000,000 fit perfectly the goal of 100,000 genomes at 30x coverage. Expect to see Sir David Bentley on the New Years Honours list next year ;-)

I’m going to continue to follow up on the other new instrument, NextSeq 500, as well as trying to find out more about the new chemistry's and imaging technologies Illumina are using, and what we might see port over to HiSeq 2500 in the 1TB upgrade later this year. SO please keep watching Core Genomics.

Incidentally I did not know it was possible to trademark a letter!

So what can you do with a HiSeq X Ten: It looks like rapid run mode is out, sort of. And that only Human genome sequencing is supported, and this only with the TruSeq Nano DNA HT kit using a 350bp fragment size; all sounds a bit limiting. In the datasheet Illumina state “HiSeq X was designed and optimised for human whole-genome sequencing; other applications and species are not supported”, I find it hard to believe people with these systems would not use them for other applications. Unfortunately nearly all of us are never going to get our hands on one, mainly because we have to buy 10.

Speed: This is where the X Ten makes huge strides over the HiSeq 2500. The new X Ten instrument completes a paired-end 150bp run in just 72 hours compared to HiSeq 2500’s PE125 in 144 hours. Per base that’s 140% speed increase, imagine a BMW R1200RT with a maximum speed of 360mph. This is what I meant by “It looks like rapid run mode is out, sort of” above, who needs a rapid run of just two lanes when a high-output run takes 3 days.

Output: The new X instrument generates 50% more clusters per run than HiSeq 2000 allowing X Ten to generate 900Gb per flowcell compared to “just” 500Gb on a 2000. It also looks like the new nanowell flowcells might allow clustering to completion within a well using all the available "real-estate" in the well to maximise signal from each cluster. Who knows if this also means the length of fragment will be less of an issue than previously, if we can make fragment libraries of 100bp and sequence those on HiSeq X Ten physical coverage of the genome goes through the roof.

The rest of the news (well some of it anyway):
  • GenomeWeb: coverage as we've come to expect.
  • Nature news: mention 30x as the gold standard, not if you're looking at Cancer genomes. Nature quotes Chad Nusbaum at The Broad saying X Ten is “like the high-stakes poker table: if  you're playing $200 a chip, people who can't afford those chips don't care.” I'm not sure about everyone else but I certainly care. Suddenly there are a group of researchers who can do the same science as me for three fold less or more, maybe I've been too blind to the fact I run a lab where only a relatively small group of people can access; so being on the other side of the fence feels uncomfortable? Nature news also asks questions about what BGI might do.
  • Opinionomics: Mick Watson says "Really, though, this blows the Ion Proton out of the water."
  • Dan Kobolt at MassGenomics: points out the same catch as I did, only Human Genomes!
  • The Scientist
  • Glenn Lockwood: A Computational Perspective, "the cost of being able to store four weeks worth of HiSeq X 10 output will run about $60,000 in capital" & "keeping up with a $10 million HiSeq X 10 system may require a compute system costing on the order of $450,000 in capital up front" are sentences to make you weep!
  • NextGenSeek blog
  • Matthew Herper at Forbes
  • Genohub: can help you find NGS service provision, perhaps even on an X Ten?

1 comment:

  1. You can now order 1 lane of HiSeq X Ten sequencing and library prep services for $1,550: