Tag Archives: sequencing

Illumina, the major player in high-throughput sequencing these days, have announced the newest version of their second generation sequencing platform, the HiSeq 2000. The machine can produce a lot more sequence, and at lower cost, than the previous Genome Analyzer II.

I’m not going into much detail about the machine: for that, see posts at Genomics Law Report, Genome Web, Genetic Future, Pathogenomics and PolITiGenomics. What I really care about is what this machine implies for the future of sequencing, and specifically what we can predict about the coming 2nd verses 3rd generation sequencing battles that will be kicking off later this year.

PacBio’s 3rd generation machine, which will be arriving later this year, will have an initial throughput of around 3Gb a day, at a price of around 1.4$ per Mb in consumable costs. I don’t know the specs for Oxford NanoPore’s machine; my guess is that it will be similar, but we’ll know soon.

Compare PacBio’s capacity to the HiSeq 2000, which will produce 25 Gb per day, at claimed consumables price of $0.11 per Mb ($10 000 for a 30X genome). In short, the Illumina 2nd gen machine is going to be able to pump out much more sequence at a much higher rate than PacBio. Both will rapidly increase the power of their machines after release, but we don’t know who will push faster (Dave Dooling thinks Illumina could push the HiSeq to 450 Gb per run with existing technology).

Of course, the competition isn’t just based on pure throughput. Read length and error rates are also important; the 3rd gen machines will also have much longer read lengths than Illumina and SOLiD, and we expect that the quality of sequence will be higher as well, giving the possibility of some real Gold Standard genomes being produced from these machines, rather than the somewhat messy genomes we get from Illumina.

This all ties in to the conversation I had with the Illumina people at ASHG; Illumina think that it’ll be a good few years before 3rd Gen sequencing can catch up with their current machines. I expect that, between now and 2014 (when PacBio release v2 of their machine), the major sequencing centres will keep a combination of 2nd and 3rd gen machines. The 2nd gen machines will be used when a very large amount of low-quality sequence is required, such as for Genome-Wide Association Studies or RNA-seq. The 3rd gen machines will be used for assembling genomes, looking for copy-number variations and studying the genetics and epigenetics of non-coding and repetitive regions.

I guess what I’m trying to say is that, as exciting and cool as the single-molecule technologies of PacBio and Oxford NanoPore are, it is far too soon to announce the death of Second Gen sequencing. If Illumina continues to push its throughput as hard as it is doing now, 2nd generation machines will be widely used for a long while yet.

The future will become a bit clearer at the AGBT conference, where we should see some big announcements from PacBio, Oxford Nanopore, Complete Genomics, ABI and Illumina. Me and a host of other bloggers will be there to cover them.