bioinformatics, genomes, biology etc. "I don't mean to sound angry and cynical, but I am, so that's how it comes across"

The de-democratization of sequencing

It seems such a long time ago (in reality: 5 years) that many people hailed next-generation sequencing as the “democratization” of sequencing.  The heart of the idea was that, previously, only large genome centres could sequence genomes; but with the advent of next-generation sequencing, relatively small facilities could now sequence Gbs per day and effectively repeat the human genome project in just a few days.

Have Illumina just reversed this?  As I have written about at length, Illumina have enabled the $1000 genome – if you have $10million for the equipment and can sequence 18,000 genomes per year.  You will also need considerable informatics kit.  This is beyond the reach of many small- to medium- scale labs, many of whom invested in Illumina machines based on the idea that they would become “genome centres”.

Have we just witnessed the de-democratization of sequencing?


  1. Well..it is to be expected that this type of things follows wave patterns right? What has now just become a really expensive investment is be accessible to the pockets of any small sized company in about 3 years time… then you will have the unde-democratization of Sequencing 🙂

  2. Hmmm, I do wonder if, with the sheer scale of the data, this pattern will not be repeated and sequencing will remain in the hands of those with strong bioinformatics?

  3. Eventually, democratization happens when the analysis can be performed directly by end-users (clinicians, doctors, biologists). If you extrapolate a bit, you could see the $500 genome, the $100 genome, the ‘the sequencing is not the major cost center,damnit’-genome. And actually, I think we are in that territory already.

  4. Even now, biologists do the interpretation, not the data analysis; multiple that by the X Ten system throughput, and it takes the technology further away from “clinicians, doctors, biologists”. This is why Illumina have BaseSpace and Isaac. Eventually the machine will do the analysis, I guess.

  5. Mick, interesting – the folks I know balked at the cost of the Illumina cloud based analysis, but that may have been them. The data analysis should indeed be fully automated (AND ROBUST). If the machine will do the analysis or not depends on the ecosystem that will evolve, some industries converge on ‘stovepipe’ solutions, where vendors sell integrated solutions that provide end-to-end service. Other industries see specialization and “best of breed” products, integrated by smart buyers or integrators. Hard to tell what will happen.

  6. I would argue that if we look at the cost of sequencing, not the price, there have always been two tiers. Many smaller facilities can’t reliably compete on total costs. Now it’s just more explicit.

  7. Actually the NextSeq has brought affordable (not cheap) whole genome sequencing into the realm of individual laboratories. The HiSeq X 10 simply allows the X-Large players to go after HUMAN WHOLE GENOME sample sizes that were not affordable before, without crushing the rest of the space.

  8. Gah. See my post on NextSeq, which doesn’t make sense to me, buying 2500 data is better value and more sustainable. Buy a NextSeq and watch it gather dust. Plus neither NextSeq nor 2500 approach XTen costs.

  9. The HiSeqX 10 is only for human WGS. Small and medium-sized labs usually don’t have 18000 samples in the freezer waiting for WGS. The CAPEX for the NextSeq and the newly priced MiSeq makes genomics more affordable to small and medium labs: this is democratizing genomics. Life Tech showed the way with the Ion Torrent and the Ion Proton, and Illumina not only matched their CAPEX strategy, but offered much better data coming out of the machines.
    Bioinformatics: it is part of the science, so the expertise in preparing a sample to put into an Illumina machine cannot be disentangled to the expertise in analysing the sequences at the other end. If small labs cannot afford a supercomputer to analyse the reads, BaseSpace is *free* right now, subsidized by Illumina. This will not last forever, and BaseSpace will cost money at some point in the future, but the trend is for cloud computing to be cheaper — watch Amazon, Google, VMWare — not more expensive.

  10. The *only* thing that prevents anyone from sequencing non-human samples on an X Ten is a license; an agreement. That license can be changed, and it will be, the only question is when. So the $1000 genome is coming to ALL species, but is only accessible to those with an X Ten. Not just the $1000 genome, but the $50 RNA-Seq, $100 exome etc etc. It will all be possible but ONLY on the X Ten.

    Are you sure you want to buy a NextSeq and churn out data at 10 times the cost of a genome centre? Because everyone else will just buy data from an X Ten owner.

    BaseSpace is not a viable option for even medium-scale projects; and as you say it won’t be free forever.

    Amazon charge ~$1000 per Tb for storage, yet in our labs we get it for ~$300. So cloud has a *long* way to go to catch up with genome centres on cost of disk storage – a *long* way.

  11. Why is BaseSpace not a viable option?

  12. Have you tried using it? I mean, actually using it?

  13. What are you trying to do that you cannot do?

  14. Well, first up, you cannot upload fastq data, it has to be written there by an Illumina machine. Which means I will never use it. Never, ever.

    Secondly, the software I want isn’t on there.

    Thirdly the app structure does not give me enough control over the parameters of the software, nor over the flow control of any pipelines I set up.

    Fourthly the reference genomes I need are not on there.

    Honestly, I could go on, but do I have to?

  15. Can you write down the list of the software (and versions) you want, the workflows you need and the reference genomes you need?

  16. You work for Illumina? I already talk to Dave Townley about this. Happy to chat more formally if you want to get my requirements.

  17. Then the question remains why people buy machines, rather than buy data from someone else…

Leave a Reply

© 2017 Opiniomics

Theme by Anders NorenUp ↑