(That would make anyone care?)
If you can choose between better, faster, or cheaper – wouldn’t you take all three?
In today’s DNA processing environments there are a few market leaders. These seem to be the result of advertising budgets more than real capabilities.
In the realm of biological material sequencing, the clear market leader is Illumina Corporation. They are huge, aggressive, and very successful. They also make a fairly high quality series of sequencing by synthesis devices. They’re also expensive. At the present time, we don’t participate in this part of the technical spectrum. But there are technologies coming that seriously challenge the preeminent position of Illumina.
First, BGI out of China has a very competitive sequencing technology that has been shown superior to the Illumina systems in several key ways. The US FDA has blocked the introduction of this system in the United States, but it is widely used in Australia, New Zealand and, obviously, China. It is also making inroads in Europe. This, in turn, is acting to bring prices down in the USA. But sequencing is still expensive, relatively speaking, in the DNA process.
Next, the coming release of real, reliable, affordable sequencing by degradation devices will undoubtedly revolutionize the art. Perhaps to the degree it can be considered science. You may have heard of these devices referred to as “Nanopore chips”. This is because of the semiconductor chip they are built on with a number of holes drilled at microscopic scales, measuring a few nanometers across. The first of these devices has been promoted by Oxford Nanopore of the United Kingdom. They are backed by the British government and have produced a device that does work, but it is slow, expensive, and not terribly accurate. It’s biggest contribution is the ability to divine epigenetic variations of some molecules. This is an advanced topic that will be a future paper. But ONP doesn’t bring the price of sequencing down. It merely extends the readable portion of the genomic patterns. This is another advanced topic for later.
Once DNA is “sequenced”, which actually means read into a complex mosaic of pieces, it needs to be put into order. This is the process of “Alignment”. And this is what we’ve got that would make anyone care.
Again, Illumina has tried to position themselves in a preeminent position in this part of the DNA processing technologies. Three years ago ad doctor at Kansas City Children’s Hospital developed a method of running the “best” alignment technology on a system of Field Programmable Gate Arrays (FPGA). These are new technologies, but this was a novel approach and significantly accelerated the alignment of sequenced data. He did this to save lives in pre-natal and infant ICU cases at his practice. It was a true advancement, but it was expensive. FPGA devices are not, and cannot be, inexpensive. Illumina bought the technology, hired the doctor, and formed a subsidiary to manufacture and sell the device, called a “Dragen” at a cost of around $350,000 each. Another downside is that you need a high end computer, such as a Dell DISOD, which costs another $275,000 in order to process the data. They recommend a two-node cluster at a cost of $475,000 for optimal processing. The advertising campaign is based on the statement “When speed matters”. Their claim that this is the fastest aligner on the market is spurious, at best. The Dragen will align a fifty times oversampled exome specimen in about 2 minutes and 17 seconds, according to Illumina’s internal testing. The output of this is a file of puzzle pieces with the chromosome and location indicated. These files can be very large, but to be useful the data has to be reviewed against the reference of the human genome to determine variations that may be pathogenic. This creates a Variant Call File, or VCF. The Dragen device doesn’t do this, and to complete the VCF takes another product and another 40 minutes to complete. Our system does the alignment and generates a complete VCF output in 2 minutes and 45 seconds. We compared the alignments and variant calling between these two products and found no significant differences. Where differences were noted it was more likely the Dragen output was in error due to a process step they label as “clipping”, which ignores portions of alignments that don’t meet expectations. So, “When speed matters”, we suggest you consider our product.
Returning to the subject of sequencing by degradation for a moment. The Dragen device uses an embedded version of the Burrows Wheeler Aligner (BWA) for alignment of sequenced data. This product has been around for nearly three decades and is considered a proven veteran. It is also 30 year old technology. Our aligner has been compared directly to BWA and we found that BWA typically creates an alignment with an error rate ranging between 1% and 3% on average. Our error rate is consistently around 0.001%. We are willing to show exactly where these errors occur and discuss the technical details of correcting them. So far, BWA and their sponsor, IBM Watson, have not been willing to discuss errors with us.
So, when you compare apples-to-apples for runtime, cost of operation, accuracy, and infrastructure required for support, our system is faster than the fastest available, lower cost by an order of magnitude, and more accurate. Infrastructure costs are also several multiples lower. So, if you want to choose between better, faster, or cheaper; why not take all three.