There has been a bit of interesting activity in long read sequencing technology investment and acquisition in the past few weeks. We saw Amgen’s ~$66M (£50M) investment in Oxford Nanopore Technologies, followed a couple of weeks later by the announcement of Illumina’s $1.2B acquisition of Pacific Biosciences. Work in long read technologies has been going on for many years, so is this a sudden rush or the culmination of long-term planning?
Looking at ILMN’s acquisition of PACB, there are both obvious and subtle things to consider. Certainly, it’s a hedge against the progress being made by Oxford Nanopore, giving ILMN a credible answer to long-read data generation beyond virtual long read approaches like 10X Genomics. Whether to investors or customers, having an externally visible play in long-read single molecule sequencing is important in keeping everyone on-board and less antsy about any potential market upheaval.
Why Long Reads?
Long read data is important for several research and clinical applications. The importance for de novo sequencing is obvious, but long-read data is also very valuable for whole genome sequencing (WGS) where repetitive elements, balanced translocations, inversions, etc., can stymie short read technologies. And in metagenomic studies, populations of organisms with many regions of high homology can make it hard to determine who is who (and which genes belong to which organism) in a mix. And with these applications, having short and long read data coming from a single supplier will likely mean tight integration and the best of both worlds: great read depth at low cost and direct, long-range haplotype resolution. This may enable ILMN to hold or even gain market share as the overall market continues to expand.
Clinical Application
In the clinical world, there are places where long reads do and do not help. Cell free applications, like NIPT and liquid biopsy, rely on the presence of short fragments of DNA released into the bloodstream, so long reads offer little benefit, and may offer less value where high numbers of reads (at lower cost) are most critical. Complex genomic rearrangements in tumors can be better interrogated by long-read technologies*, though there is little on the immediate horizon that would provide clear clinical utility in drug selection. But there are some clinical applications where long reads may be very useful, such as transplantation, cellular immuno-oncology, and microbiology.
Clinical applications enabled or improved by full-length, phased HLA sequencing definitely benefit from long-read approaches. This may range from conventional transplant matching to allogeneic cell therapies, and potentially some forms of autologous cell therapies where antigen presentation is MHC mediated. And long read V(D)J sequencing should streamline and improve immune repertoire sequencing, which should be increasingly useful as immuno-oncology therapies evolve. Given all the investment and early successes in immuno-oncology (not to mention other areas like autoimmune disease), being well-positioned in long-read technologies will offer returns in years to come.
The Need for Speed
There is also another feature of long read technologies that makes them well-suited to particular clinical applications, which has to do with the difference in the nature of parallelization in “conventional” NGS (e.g., ILMN SBS and TMO “semiconductor” sequencing) vs. single molecule (PBIO SMRT and ONT nanopore sequencing). The conventional NGS approach is to sequence in parallel many clonal groups of molecules one base at a time**, so each sequence takes a step forward during each cycle of chemistry. By doing this in a massively parallel manner, a huge amount of sequencing data can be generated, but full length sequences aren’t available until the end of the run. With both PBIO SMRT and ONT nanopores, each fragment is sequenced continuously in a given well or pore before the next one begins, so full length sequences are generated shortly after the run is initiated. This, combined with the dramatically longer reads enabled means that long, uniquely identifiable sequence reads are available very quickly, and in time-critical applications this can be a key benefit. Think particularly about rapid microbial identification and detection of resistance mechanisms in life-threatening infectious diseases, and the benefit is clear. These approaches still require library prep, which can be time consuming and laborious, but at least the sequencing readout is fast enough to provide useful information in a clinically meaningful timeframe. Considering ILMN’s past successes in taking a complex technology and packaging it into a simplified workflow, PBIO’s systems may undergo a rapid evolution that makes them serious competition here in a way their SBS technology might not.
A Seat at the Table?
The other recent news in the NGS space was Amgen’s investment in Oxford Nanopore, which Amgen stated has to do with target discovery and validation (rather than a diagnostics play). Given Amgen’s acquisition of deCODE genetics, this rings true for the most part, not to say that companion diagnostics might not emerge from their work. But it seems to align more with Amgen’s investment in building an advantage in target discovery, and getting a seat at the table further upstream of others, including access to genomic technologies. Taking a look at Amgen’s therapeutic modalities and its early stage pipeline, there are areas where long read sequencing is very valuable, particularly around immune repertoire sequencing and immuno-oncology. If ILMN takes some time bringing next-gen PBIO systems to market, then having deep access to ONT’s technological bench may help accelerate some of their discovery efforts, or at least ensure they have good access to long-read capabilities. But as ILMN tightens its grip on the market and moves deeper into clinical applications, it will be interesting to see if Amgen’s investment offers them access to technology in a way that gives them a competitive advantage, or just a share in an ultimately commoditized capability.
-Adam Lowe
Notes:
*See, for example, Norris et al. (2016) Nanopore sequencing detects structural variants in cancer. Cancer Biol Ther 17(3): 246-253 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4848001/
**In Ion Torrent semiconductor sequencing, a sequence will advance several bases whenever a homopolymer stretch is present in the template