Everything Wrong With: "Illumina: The Measurement Monopoly" Part 1/8
Century of Biology has 10x the subscriptions I do… I could lie and say I’m not bitter about this. But well…I am, please subscribe.
As an exercise in catharsis I’ve decided to go through Elliot’s relatively recent post on Illumina, and moan about it.
Starting with the first line:
“No technology has ever improved more rapidly than DNA sequencing”
No, this is wrong. You’re wrong Elliot1! Over the last 16 years improvement in sequencing has been at best incremental.
Illumina have been meandering along making solid slow progress since they acquired Solexa. It’s fine, but it’s nothing compared to other industries, in particular the most obvious comparison of the semiconductor industry:
Look at that! Every 10 years they’re fitting 10x the transistors on the same piece of silicon over a 50 year period!
How are we doing in sequencing? We’ve gone from 750,000 per sq. mm on the Genome Analyzer to about 8 million on Illumina’s super-resolution instruments, a 10x improvement, over 16 years. And at least to me, it kind of looks like things are stagnating.
In comparison the semi-conductor industry was moving even faster than 10x/10 years in the 1990s increasing density by ~15x every tens years.
And it’s been way way harder in the semiconductor industry too. Just for the lasers they build massive room sized lasers where they fire multiple lasers at falling droplets of tin. It’s crazy and it costs billions of dollars.
Sure, there was a big leap forward with the launch of the first Next-gen sequencing instruments. The 454 GS20 generated 200K reads. Capillary sequencers ~100. But is capillary really the right comparison? Array’s were providing readout of 100K+ probes.
Early in the development of new approaches big technological leaps like the GS202 are relatively common.
Look at this early memory IC on top of the (then state of the art) core memory device:
The first commercial DRAM IC was 1000 kilobits. Probably 100 to 1000x the density of core memory.
What’s harder is sustaining this pace of development over decades. Something the semiconductor world has accomplished. But where sequencing so far, has failed.
Illumina Sequencers Are Boring Now
Illumina’s best selling sequencer was last updated 11 years ago. Elsewhere in their lineup they keep releasing new instruments which are pretty much the same as the old ones just bigger.
It’s not old bad news, Illumina have made some legitimate advances since the Solexa acquisition. In particular a few new key technologies have been introduced:
Patterned Flowcells and ExAmp.
TDI imaging.
BUT THEIR BEST SELLING INSTRUMENT IS STILL AN 11 YEAR OLD COST OPTIMIZED GENOME ANALYZER.
These are all neat incremental developments incorporating advances from other fields in many cases. But for the most part Illumina have only pushed forward when forced to by the market… and with the exception of the MiSeq (and the Iontorrent PGM), nobody has really released anything competitive.
So they’ve been slowly pushing out these incremental enhancements (which are all really great, but mostly unexciting!). To get users to buy new instruments and grow the market a bit, but doesn’t really shake things up.
Just Throw More Cameras In The Box
Do you remember the 90s? I kind of remember them… just about. In the 90s computers were getting faster and faster. 60Mhz, 100Mhz, 500, 1000! 2GHz! Seemed like every 6 months processor speeds would increase. Then it just… kind of stopped.
Instead they started giving us more cores. Manufacturers started hitting fundamental limits and couldn’t easily increase processor speed any more. So they figured… well just throw more cores on the die.
Illumina have been doing the same thing. The Genome Analyzer used a single camera. The HiSeq 2 complete optical systems3. And recent instruments like the NextSeq 550 cram 6 complete optical systems into a small bench top box.
Just throw more cameras in the box and make the flowcell bigger:
I’ve no idea how many cameras are in the NovaSeq X. But the flowcells are huge and it takes two of them!
Again… this is… fine. But not the exciting technological explosion that Elliot’s first sentence makes the current sequencing market out to be.
Competition Is A Trainwreck
Why am I only talking about Illumina? Well… because there still isn’t much in the way of effective competition. Illumina’s basic sequencing patents have finally started to expire. The response from the competition has been to target the least cost sensitive part of the market (the mid-range) with instruments and consumables which are only a little bit cheaper than Illumina.
Surprisingly customers have been largely reluctant to buy $100K+ instruments from companies that may not exist in a year. Long read platforms are neat, but too expensive… and so most users stick with Illumina…
In Summary….
So… in view of how boring the sequencing industry currently is, I would rewrite Elliot’s first sentence as follows:
“DNA sequencing improved a lot with the launch of next-gen sequencing in the 2000s. But it’s advancing relatively slowly at the moment and we continue to hope for some true disruption in the market.”
And this is why Elliot has 10x the subscribers I do.
If you enjoy overly pedantic, often painfully technically detailed content. Consider subscribing.
Well… maybe you could be right or wrong. But that’s even worse! Not Even Wrong!
From 454 which ultimately failed to survive!
Two sets of two cameras.