Although Illumina’s two new NGS platforms overshadowed their press conference at the J.P. Morgan Healthcare Conference, there were other interesting tidbits as well. When one gets past the hype and somewhat-unrealistic calculations of per-genome sequencing costs, it’s good to remember that Illumina’s current platforms already have ~80% of the next-gen sequencing market and a similar command of the genotyping array market. They’re investing heavily on the informatics side as well. Here are some of the highlights.
HiSeq and MiSeq Instruments
I’m guessing that the majority of those customers don’t have $10 million handy to plunk down on a HiSeqX Ten cluster. Thus, it seems likely that the MiSeq and HiSeq will remain workhorse platforms for much of the research community. In the last quarter, Illumina booked around 400 instruments, and 300 of those were MiSeqs. Half of the overall demand came from non-academic customers, and Japan was second only to the U.S. in terms of growth. Illumina expects MiSeq output to reach 15 Gbp per run this year, which is enough for 2-3 exomes or a lot of more targeted sequencing.
Longer Reads in Rapid Run
In the second quarter of 2014, Illumina plans to roll out longer paired-end reads (2×250 bp) for the HiSeq2500 rapid mode. As a bioinformatician, I’m a huge fan of (slightly) longer reads with faster turnaround times. As a manager, however, I can’t ignore the cost of said capability. Running the HiSeq2500 in rapid mode is already expensive, reagents-wise, and when you generate longer reads, you need more kits per run.
There’s another reason 2×250 bp reads might be wasteful: if the typical DNA fragment size is 250-300 bp, most fragments will be sequenced twice (once from each end). That was great in the 3730 days, but many NGS pipelines soft-clip the overlapping portions of paired-end reads from the same fragment. In other words, they only use the information from one read. Then again, speed and accuracy are likely higher priorities than cost in clinical settings, where the rapid run is more likely to be employed.
Speaking of which, the HiSeq2500 will be submitted for FDA clearance this year. Well, that’s the tagline. Specifically, they’ll submit the use of the 2500 on a specific genetic test for FDA approval, and build out from there.
Illumina’s SNP Arrays
There’s also the SNP array business, which has been booming for a long time. We like to talk about 10,000 or 50,000 exomes, but chip-based GWAS cohorts typed by SNP arrays often number hundreds of thousands of samples. The exome chip, for better or worse, saw rapid and widespread adoption over the last couple of years. They’re cheap, they’re easier to analyze, and they require simpler consenting procedures than exome or genome sequencing. These arrays are selling like hotcakes.
During the J.P. Morgan conference, Jay Flatley dropped this interesting line: they have seen no decrease in demand from their biggest SNP array customer, 23andMe, despite the FDA’s warning letter. I knew that 23andMe used Illumina genotyping arrays (their own customized version), but I was surprised to learn that they’re the biggest customer. It’s certainly interesting to learn that the personal genotyping service business remains strong even though 23andMe can’t currently market the health or medical benefits.
Then again, since it costs only $99 to get genotyped, even the ancestry stuff is fascinating.
Illumina’s BaseSpace Informatics
One other area of movement is BaseSpace, Illumina’s cloud computing platform for NGS informatics. They’ve connected over 2,000 instruments to it already, and registered 12,000 user accounts. There are 25 BaseSpace “apps” available now, and that number is expected to double by the end of the year. Supposedly, the amount of data already in Illumina’s cloud already exceeds the size of the NCBI short read archive (SRA, not to be confused with dbGaP).
Also, there’s a “private” version coming out: Illumina BaseSpace Onsite, which contains all of the reporting without being on the cloud. Yet another development aimed at the clinical market.
Sure, BaseSpace is growing, and it’s reassuring to see Illumina investing heavily in informatics (remember when all they had was ELAND?). Bioinformatics already represents a slower and more expensive prospect than sequencing. That bottleneck will only grow worse when HiSeqX Ten installations start churning out 20,000 whole genomes per year. If they can find that many samples, that is.