The next chapter of a long read
This article is entirely personal opinion and does not represent the official position of any entity.
On November 1st, Illumina announced it is acquiring Pacific Biosciences for $1.2 billion. For Illumina, the decision is very smart, as it must be aware that as all sequencing methods improve, long reads will become increasingly attractive and eventually essential. For Pacbio, it was a necessary move. With a 100% premium on the prior share price, it is a respectable exit.
My life has been immensely enriched by working with the scientists at Pacbio and with the technology they built. Almost two years ago to the day, my Halloween costume was Captain Genomics, armed with the shield of Circular Consensus (a technology that aged very well) and the scaffold of BioNano.
So I’m prepared to put my money where my mouth (and questionable fashion sense) is. I feel compelled to write this article as something between a critique and a letter of respect.
The defining characteristic of Pacbio was that it generated such immense scientific value to the community despite lacking the ability to capitalize on more than a fraction of its contribution. Even more, their scientists freely and happily built open-source methods which maximized the community value and the company supported this fully. Do not let anyone tell you this was a weakness, it was one of their greatest strengths.
The scientific fruits of Pacbio are so ubiquitous, though I think few realize it. How many high quality reference genomes have been produced just in the last year. I think Nicholas Hill, a very smart and qualified intern, assembled more plant reference genomes in a summer than existed in the world at the start of my PhD. Assemble the transcriptome too? That’s one assay away.
Similarly, if you take the time to stop and appreciate Genome in a Bottle (which you should, every day), you should understand that a huge amount of the scientific lift which enables confident calls is the value provided by Pacbio as a highly mappable technology with a fundamentally different error mode from Illumina. I don’t think GIAB could reliably be done at all without Pacbio. This is true to an even greater extent in structural variation.
However, the greatest commercial leverage for these applications was not captured (or capturable) by Pacbio. Reference genome assemblies enable at-scale resequencing of populations. Genome in a Bottle’s greatest immediate effect was to spur incredibly rapid improvement of informatics approaches on short-read methods. I love using long read data, but one of the first things Samantha Zarate and I used the high quality structural variant sets for was to improve the accuracy of Parliament2, an ensemble method of… short read callers (and nobody at Pacbio ever yelled at us for this).
This isn’t Pacbio’s fault. This was the niche they could occupy effectively. It turned out to be really, really hard to scale the cost of single molecule methods. And though they did achieve a solid cost trajectory, it wasn’t quite enough to achieve full escape velocity. My point is that they were aware of their position, but I never saw their scientists truly begrudge it. They may sometimes have doubted the sustainability of their business as an independent entity, but I don’t think they doubted the value of their science.
Along the way, they innovated many new informatics methods, like FALCON, which they freely open-sourced, published, and distributed. If ever anyone tells you this was a mistake, push back hard. If they had not done this, they would not have achieved commercial traction and would have died years ago.
The scientific community loves Pacbio. They have advocates in every corner of the globe pushing for adoption of the technology. Not that I don’t have complaints about every implementation decision. On the science side, Pacbio did community engagement right. It was obvious that they were genuinely driven by the science. Illumina would do well to preserve this spirit.
OK, so what does this mean for the field. Let’s talk about the best world first. In the best world, Illumina and Pacbio combine the best of both their worlds. Illumina leverages its fast resources to supercharge Pacbio R&D, making long reads more affordable and rolling out powerful hybrid sequencing applications. At the same time, Oxford Nanopore fires on all cylinders, demonstrating strong market position in unique and defensible niches (like field sequencing) and effectively competing on cost. From a position of strength on all sides, we have a race to the top. I want to live in that world.
In the worst world, alternatives to Pacbio have difficulty scaling into the market. Illumina is not as pressured to invest, as they have a leading market solution in both niches. Pacbio remains a strong solution to build resources for at-scale application of short read methods.
There are many other scenarios — ones where either way Illumina invests aggressively, realizing that maximizing value of a “premium” technology lets them move more volume of a higher value product.
My advice to Illumina: recognize those parts of Pacbio which were its strengths — their open-source development, comfort with partners developing informatics ecosystems outside of company control, and the way they had a collaboration-first philosophy with experts in the field.
Finally, I want to thank all of the former Pacbio scientists I have had the honor to work with directly: Andrey Kislyuk, Marcus Kinsella, Maria Nattestad (the internship counts), Jason Chin, and David Alexander. And the many still at Pacbio I have had the honor to collaborate with.
I suppose this sounds like a goodbye story, but I hope that fundamentally it is not. I am absolutely certain that the future of is the ubiquitous deployment of cheap long-read methods. I hope that this is not the end of the story, but is instead turning the chapter of what will be a very good and incredibly long read.