Automated DNA Sequencing Industry: Far From Commoditization
Contributed by: Alejandro Gutierrez
Automated DNA sequencing ramped up in the late 90s with the introduction of the capillary electrophoresis (CE) instruments developed by Applied Biosystems. This platform played a key role in the race to sequence the first human genome, and retains a tenuous hold as the gold standard for de novo sequencing, particularly of larger genomes (e.g., mammals). CE represents the first-generation of automated sequencing technology, whereas the short-read, high-throughput platforms commercialized by 454 Life Sciences, Illumina and Life Technologies represent the second generation. These instruments have been on the market since the mid-2000s and their performance has improved rapidly without fundamental changes in the underlying technologies. The first third-generation, long-read / high-throughput platform is already on the market and others are expected soon.
As these ongoing, overlapping waves of technological innovation hit the market, there is very limited evidence that instrument commoditization is underway. Several relevant indicators, examined below, evince an industry with significant unmet customer needs, integrated competitors, differentiated products, and relatively high barriers to entry – none of which are signs of an industry on the verge of commoditization.
What the customer wants
When automated sequencing first emerged, researchers involved in the sequencing of the first human genome faced an enormous task and there was a large, unmet need for faster and cheaper approaches. CE platforms were a great improvement over manual methods, but there remained significant room for additional gains. Second-generation platforms introduced a variety of new technical paradigms that allowed another jump in performance. For certain applications, these platforms have begun to approach the desired performance level of some customers. But for most applications, performance by second-generation instruments, and even the emerging third-generation platforms, still does not meet the needs of the market.
The main limitation of CE technology is its low throughput and very high cost. The main limitations of second-generation technologies are their short read-lengths (relative to CE) and still high (although constantly falling) cost. Accuracy is also an issue, although it can be overcome with increased coverage, at a higher cost.
Read length. With the exception of very short molecules, such as small RNAs, the length of most molecules to be sequenced exceeds the average read length capabilities available today. Different techniques have been developed to overcome this shortcoming. The fundamental approach to genome sequencing, DNA fragmentation and (re)assembly, exists because technologies cannot read entire DNA molecules, or even relatively small fragments. The shortest human chromosome is roughly fifty million bases long – compare that with today’s maximum read length of ~1,000 bases. Paired-end and mate-pair reads are other techniques used to overcome short read lengths. But all of these techniques have drawbacks, and are not yet capable of properly addressing highly repetitive areas of the genome. The very fact that researchers and manufacturers are striving to find ways of stretching current technologies beyond their optimal set of applications is evidence of unmet need. Longer read lengths should enable complete genome coverage, reduce complexity of assembly, enable new applications, and improve cost-effectiveness.
Cost per base. Despite the impressive drop in costs for sequencing, costs remain too high for a large swath of potential applications. The expressed goal of a $1,000 genome, and the claims by many firms of a $100 genome, are a recognition of the untapped markets that could open up as costs fall further. Even for existing applications, which are largely in research, ever-lower costs will remain a welcome improvement.
Accuracy. The accuracy goal set by the “$1,000 Genome Prize” is one error per 100,000 bases. To achieve such high accuracy today, platforms rely on repetitive sequencing of the same base. Assuming errors are not systematic, enough reads of the same base will yield a consensus estimate with a higher certainty of accuracy. The number of repetitions required varies by platform, but can often range between 20-100. More repetitions means more cost. Not all sequencing applications require high accuracy, but clinical applications – including those for diagnostic and treatment purposes, do. Given that clinical applications represent a very large potential market segment, cost-effective accuracy remains a significant unmet need. According to Elaine Mardis, a geneticist at Washington University, as quoted recently in Forbes, the new HiSeq 2000 Illumina machine is perfect for researchers like herself who need to sequence a lot of DNA. But it’s still not even close to being good enough for medical testing laboratories like Quest Diagnostics and LabCorp, which run huge volumes of tests and will need easy-to-use machines to make mass-market genetic scanning practical.
Instrument cost and complexity. Aside from the variable cost of sequencing, the cost of instruments and the complexity of running them represent a large barrier to adoption for entire segments of potential users. Smaller research and clinical laboratories, for example, rely on commercial sequencing providers or academic core labs for their sequencing because they cannot justify (or afford) the expense of buying and setting up their own instruments. The trade-off is a lack of control and convenience. There is an unmet need for smaller, more flexible instruments that can be priced attractively enough for the lower tiers of the market and allow these users to do their sequencing in-house rather than outsourcing it. 454 is one firm that has recognized this and is in the process of launching a “GS Junior” platform. But this instrument, expected to sell for $100,000, will only be the opening salvo in the push to expand the sequencing market. Beyond instruments, service providers will also help expand this market with lower-priced offerings.
As long as critical customer needs remain unmet, existing and new instrument developers will push the boundaries of technology to seek the necessary solutions. Rather than converging toward a standard, new technical approaches and paradigms will continue to be tested, as is evident from the emerging population of third-generation technologies.
Proprietary and integrated
In commoditized technologies, technological innovation to push the boundaries of performance no longer matters as much because customers care primarily about cost. Thus, components are typically modular and architectures are standardized (read more on this framework in The Innovator’s Solution). The role of the OEM is to assemble from the best, lowest-cost components available.
Sequencing instruments today are far from being modular or standardized. Each platform on the market, and each platform being developed, is characterized by a highly complex, interdependent and integrated design. This integration extends from the sample preparation process, which is unique to the needs of the platform, to the platform itself, and ultimately to the informatics engine that processes the sequencing outputs. It would not be easy (or perhaps even possible), in other words, to use the sample preparation process of the 454 GS FLX, sequence the outputs in the Illumina GAII, and then process the data using Helicos’ HeliScope Analysis Engine.
Inside each platform, companies are using custom-developed and off-the shelf components, often leveraging technologies developed for other industries and applications. The types of components used in each platform, and the way in which they are integrated, are specific to each manufacturer’s approach and design. The reagents that are used by the sequencing instrument are also proprietary, unique and optimized to the individual platforms. Even though all second generation sequencers rely on enzymes and nucleotides, these materials are engineered to provide the necessary reactions specific to each platform.
Differentiation and specialization
While instrument manufacturers will tend to play up the broad spectrum of applications that their instruments can tackle, the reality is that each platform today is best suited only to a range of applications. Applications beyond this “sweet spot” are possible, but they come with increasing trade-offs, usually in cost or quality. In many cases, there is another platform, or even another technology (not sequencing), that is better suited for the job. In some cases, there is a set of applications that is not properly addressed by existing second-generation offerings – the most salient example being applications that demand high-throughput and long reads.
Although specialization means that there is no ideal one-size-fits-all solution for all sequencing applications, it also means that not all platforms compete head-to-head. Even as second-generation technologies have entered the market, first-generation CE platforms still find their uses. As third-generation platforms begin to appear, they will not directly replace second generation platforms in applications where short-reads and high-throughput are needed. Evidence of the recognition of these platforms’ different capabilities can be seen in the deployment of diversified portfolios of instruments in the larger academic and research centers.
An important question is whether the specialization seen today is a consequence of technical design choices made by the manufacturers, or a conscious strategic choice to pursue a particular set of applications. The latter could imply a trend toward best-in-class platforms that are narrow in application and highly differentiated. On the other hand, if specialization has been a consequence more than a goal, companies could end up developing platforms that have a broad range of applications even if they are not the best at any of those applications. A herd mentality could lead to a battle of one-size-fits all solutions.
Both differentiated and all-in-one product development strategies may succeed in different market segments, based on customers’ needs for different applications, and the actual trade-offs in cost and performance between the different platforms. Certain applications, however, may prove to have much larger markets, so not all strategies will yield the same payoff. Clinical diagnostics, for example, could prove to be a very large opportunity where a differentiated, specialized platform would have the edge over an all-in-one instrument.
The emergence of all-in-one platforms that meet the needs of a specific market segment could be an indicator of the start of commoditization. To some extent, the second-generation platforms offered by Illumina and Life Technologies are probably the closest to meeting the performance requirements for their sweet-spot, short-read applications. Although these platforms differ along several technical and performance attributes, their race for higher throughput represents, to a large extent, a descent into cost-driven competition. Outside of this example, however, differentiation and specialization among existing and forthcoming platforms remains the norm.
Barriers to entry
At first glance it may appear that entering the sequencing instrument market is not a difficult proposition – assuming a core technology has been identified. The list of start-ups is fairly long and new companies appear to emerge every few months. In addition to the six commercial players, at least nine emergent players are vying to get in:
|Life Technologies||Pacific Biosciences|
|Illumina||Visigen (acquired by Life Technologies)|
|454 Life Sciences (Roche)||Oxford Nanopore Technologies (funded in part by Illumina)|
|Polonator (Dover; Danaher)||ZS Genetics|
|Complete Genomics||Halcyon Molecular|
|Helicos||Ion Torrent Systems|
Notwithstanding the number of current and potential competitors, the capital required to reach the market is significant and hard to come by. Initial funding needs are not as significant and may be easier to raise, but at the stage of scale-up and commercialization, capital requirements can be daunting. In addition, the current climate for raising private and public funding is constrained, and in the life sciences tools sector investors hesitate further given the paucity of exit strategies (see, for example, this Genome Web Daily News article).
|Pacific Biosciences||$250M+||Product not yet launched|
|Complete Genomics||$90M+||Product launched|
|Helicos||$67M+?||Product launched, public|
|ONT||~$80M||Product not yet launched|
|Illumina (Solexa)||$650M||Price for Solexa acquisition (all stock)|
|Roche (454)||$150M||Price for 454 acquisition|
|VisiGen||$20M+||Price for acquisition by LIFE|
|Avantome||$25M plus up to $35M in milestones||Price paid by Illumina for acquisition|
Another barrier to entry is the established sales and service infrastructure of Illumina, Life Technologies and Roche MDx, all of whom can leverage it over a number of different product lines, giving them cost and reach advantages. Emerging firms must develop their own infrastructure (which will, initially at least, be dedicated to only one product), or establish partnerships (e.g., Dover’s tie-up with Azco Biotech to sell the Polonator.)
Reputation and installed base are other barriers to entry, especially among existing customers. Early adopters may be more willing to try new technologies, particularly on a limited-risk basis, but the broader market, particularly customers that are more cost-sensitive, may prove more reluctant to spend upwards of $500K (plus the additional commitment to infrastructure and training) on untried firms – especially if they have an installed base of other instruments. (There are counterarguments to the “installed base” barrier. As an analogy, look at the airline industry, where many fleets include both Airbus and Boeing jets in the same category, e.g., A320 and 737. That said, consider also Southwest Airlines, whose fleet is entirely made up of Boeing 737s.)
The sequencing industry is not yet on an obvious path to commoditization. Most signs point to a continued and likely extended period of performance-based competition with opportunities for differentiated product strategies and significant profit potential. Nevertheless, it is fair to consider what the likely path and pace of commoditization could be. There are numerous theories of the factors that drive the pace of commoditization. Some of these include the rate of overall economic expansion, growth in demand, production capacity and the emergence of overcapacity, the pace of technological change and performance overshooting relative to customer needs, intensity of competition and managerial obsession with beating competitors and stealing share, and consolidation of the customer base (see this article, for example).
Some of these factors are extrinsic to the sequencer manufacturers, while others are within their control. The rapid pace of technological innovation would point toward an earlier overshooting of performance. If current specialization is more accidental than strategic, it may indicate a mindset among competitors to move towards head-to-head, cost-driven competition. The growing concerns over overwhelming data output, and the inability of informatics tools and resources to keep up, might be a temporary hiccup or a sign that demand could fall below capacity. But for each of these instances, there are offsetting circumstances, many of them outlined above, that make it difficult to conclude that commoditization is imminent or will necessarily be pervasive.