Opiniomics

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

Complete Genomics Revolocity and the future of genome sequencing

I’m a bit late to this, please accept my apologies – I’ve been on holiday!  If you are into genome sequencing then you cannot have missed the announcement of the Revolocity system from Complete Genomics.  So what does this all mean?  Let’s put it all in context.

Revolocity

Key features appear to be:

  • The system costs $12M and occupies 1500 square feet.
  • Automated sequencing from biological samples, including blood and saliva
  • Capable of 10,000 WGS at 50X coverage.  To be increased to 30,000 WGS (no hint of how, though I expect read length to increase)
  • Capable of exomes
  • Has necessary compute and bioinformatics built in (possibly through BGI’s cloud service?)
  • 2x28bp reads apparently covering 96% of the human genome
  • 8 day sequencing time

It’s not just the sequencer, though….

It is a good thing that Illumina have a competitor in this space, as competition drives innovation and brings the cost of sequencing down.  The Revolocity is obviously a competitor for Illumina’s HiSeq X Ten and X Five systems.  However, Illumina have a few key advantages:

  • A history of supporting machines in the field (Complete have always been a sequence-as-a-service model)
  • An international support network of field engineers
  • An international distribution network for spare parts
  • A global production line for flowcells and reagents
  • The administrative support required for selling and supporting all of those things
  • A global network of scientists working with Illumina data for many years
  • A huge ecosystem of free and commercial bioinformatics software

CG will need to reproduce all of the above to genuinely compete.  A key piece of missing knowledge about Revolocity is: how often does it fail and how quickly will CG be able to fix it?  Illumina are not perfect in this, but they are a known entity.  Can Revolocity match or even beat Illumina’s “up time”?

Comparison to HiSeq X

CG have the advantage of automated sample prep from blood/saliva; 50X coverage; bioinformatics apparently included; exome enabled.

The HiSeq X also has automation tie-ins with the NeoPrep and their collaboration with Hamilton; they also have BaseSpace for bioinformatics needs.  So not too different from Revolocity.  We need more details to do a true comparison.

The HiSeq X wins on sequencing time (3 days) and read length (2x150bp).  The latter is key.  Despite CG claiming that they will cover 96% of the genome, figure 1 from this paper suggests that 56bp will get you about 92% of the human genome, whereas 300bp (HiSeq X) gets you about 97.5%

How will Illumina respond?

I have no doubt whatsoever that Illumina will reduce the cost of human genomes further; they will do this in January, at the JP Morgan conference.  I’m guessing a figure closer to $600-800 per genome.

I am less sure, but I think they may also open up the system to human exomes and RNA-Seq.

Cost?

People have said Revolocity hits the $1000 genome mark, but I have yet to see evidence.  If they can, they are doing incredibly well.  However, we need to see more than press releases. Having been involved in the set-up of a HiSeq X system at Edinburgh Genomics, I can confidently say that the $1200 genome is genuinely achievable – by that I mean you provide us with DNA, we sequence to 30X, align, variant call and provide you with FASTQ, BAM/CRAM and GVCF, and it costs around $1200/£800.  If you are interested in Edinburgh Genomics’ service at that price, please get in touch!

Edit 14/06/2015

Revolocity cost was announced at ESHG:

The Future

I have spoken to a few people and we all agree – Illumina will remain market leader for at least 18-24 months.  That’s a long time in sequencing!  However, it is the PromethION that we think will disrupt Illumina, not Revolocity.  If the PromethION can first match and then surpass MinION’s quality levels, tied to massive increases in throughput, then it will become a serious competitor to Illumina.  Why will this not happen sooner?  It will take time for the new platform to develop, and yet more time for the market to react (a market heavily invested in Illumina).  If/when PromethION does become a threat, it will be fascinating to see how Illumina respond.

7 Comments

  1. I am guessing that the fragment insert size and paired end sequencing resolves the discrepancy between mappability rates from the figure you link to? 28 bp does sound rather disappointing though.

  2. The Genome Research paper looks at kmer uniqueness, and in the human genome, about 92% of 50-mers occur only once. There will be some difference between mappability of 2x28bp reads from a 300bp insert and the uniqeness of 50-mers, but I don’t think it will be that big.

  3. Well, it’s the mappability of two 28-mers constrained within some distance isn’t it? Which is presumably quite a lot more and somewhere between k >= 56 and k <= ~300. A few people have looked at this in various publications but I can't see a reference that shows exactly the metric we are after.

  4. How accurate will PromethION be in comparison to Illumina sequencers?

  5. Isn’t Illumina a major investor in ONP? I wonder how that will play out…

  6. nm… not anymore I suppose…

Leave a Reply

© 2017 Opiniomics

Theme by Anders NorenUp ↑