Tuesday, 13 January 2015

Illumina's new sequencers - my initial thoughts

Illumina continue to push hard on making DNA sequencing cheaper and easier to access for everyone. The announcements at last night's JP Morgan included a smaller X Ten, HiSeq 3000/4000 and NextSeq 550. I'm not going into much detail about the instruments, you can get the specs on Illumina's website, but I will talk about the impact the next step in Genome Analyser evolution.

Briefly the XTen is joined by the X Five System, half the price and only 9,000 genomes a year. The NextSeq 550 allows labs to also scan Illumina microarrays (remember those), but the focus is clearly cytogenetics and PGD - the system is not capable of running HT12's for instance.

HiSeq 3000/4000: The latest update for HiSeq is another big step forward, and somewhat surprising as it comes only a year after the release of V4 chemistry on the 2500. The patterned flowcells are here. Interestingly there was no mention of upgrades from 2500 to 4000. The 4000 is the machine I'd be wanting in my lab and the specs are pretty exciting: 1.5Tb in three das on PE150bp, 5 billion clusters (the Ga was lucky to get 1M per lane when first released), and still with high quality and ≥ 75% bases above Q30. 12 Human Genomes or 180 Exomes or 250 RNA-seq per run - we'll be busy making libraries!

ctDNA: Illumina also presented their circulating tumour DNA R&D program with kits coming in 2015 something we'll be watching closely.

I'll post again in the next day or two once I've distilled my thoughts on the impact for labs like mine.


  1. Interesting that Illumina have partnered with Lockheed Martin. Clearly the intention is to reduce price through economies of scale. Also interesting the CEO claims that 10% of applications require long read technology although no reference was provided to support the claim, and really, who can possibly state that right now with our still basic understanding. ....but regardless, Illumina chemistry is not capable of direct epigenomics which has to be the next objective and will make a massive difference.

  2. oh and longer reads mean cheaper bioinformatics.

    1. Not at 85% accuracy. There is a computational price to pay for PacBs higher error rate. If you get longer reads at 1% error then yes, the computational burden drops.

    2. ok agreed for pacbio but pacbio and illumina are fluorescence and chip based so removing these restrictions will still reduce bioinformatics burden for higher error rate long reads.