This article has been updated from a version posted Sept. 27 to include comments from a Knome early-access user.
Knome said this week that it is taking orders for a system that bundles its genome interpretation software with a compute cluster to create a "plug and play" system for whole-genome analysis.
The system, called the KnoSys 100, will begin shipping in the fourth quarter and will have a starting price of $125,000. It accepts data from Illumina, Life Technologies, and Complete Genomics sequencers.
The company has launched an early-access program for the KnoSys 100 and said that so far "more than a dozen" medical institutions are piloting it.
Knome is positioning the system as an all-in-one informatics station for labs that are installing next-gen sequencers and lack in-house computational resources but don't want to send their data over the cloud for analysis. It marks a departure from the company's other products — KnomeBase and KnomeDiscovery — which both require customers to send sequence data to the company for analysis.
The system actually grew out of discussions Knome had over the summer with early-access customers for KnomeClinic, a software suite it is developing for clinical interpretation of human genomes (BI 6/15/2012).
Martin Tolar, Knome's CEO, said that these early-access users — clinical research groups who are "preparing for the application of whole-genome sequencing and analysis in patient care" — had a number of recommendations, but the primary one was that "they wanted a solution that was within the four walls of the institution for privacy and regulatory reasons."
Initially, he said, "we were considering having enterprise software that would be installed at each of the institutions, but it became very clear that not everybody had the hardware and the capabilities to run such a complicated system. And also, [we decided that] if we can optimize the hardware for the software that we've built, it's going to be much more effective and efficient. So that's why we decided to put it all in one box."
Tolar noted that the company views the KnoSys 100 as "a stepping stone to a network solution and a cloud environment" for its products, but for now, "we heard over and over they want something that will sit right next to the desktop sequencer."
The KnoSys 100's throughput is in line with that of current sequencers. The company claims that it can process on average one genome per day, which should allow most labs to keep up with the data coming from next-gen sequencers.
Eventually, Knome will release an enterprise-wide solution for clinics that are doing thousands of genomes, said Jonas Lee, senior vice president of marketing for Knome. "But right now what we saw is that the research labs, the translational labs, [want] something that is plug and play and can do roughly what a genome sequencer can do — roughly 30 a month — and that's what we put together as a bridge to full-scale rollout, where the technology and the interface is the same, but the scale is much more compact."
The KnoSys 100 includes four 2.4 GHz 8-core Intel Xeon E5-2665 processors and 18 TB to 54 TB of useable disk storage.
Lee said that 54 terabytes of storage should keep most labs "satisfied for many months" since the system's analysis pipeline — based on its kGaP interpretation engine — "distills" each genome to about 50 gigabytes. The company plans to eventually launch a series of larger-scale KnoSys boxes for research groups processing hundreds or thousands of genomes per month.
The core of the system is the kGAP engine, which processes variant calls into standardized and annotated data sets based on Knome's curated knowledge base of reference information.
Lee said that the system allows researchers to create "comparison databases" for specific groups of genomes, "so you can specify, 'I want a comparison file for these five or 50 or 500 genomes to be created, and our system will generate that so you can quickly and flexibly query multiple genomes together."
While Tolar and Lee stressed that KnoSys is intended for clinical research, and not clinical practice, the company is putting the pieces in place to nudge its software further downstream with an eye toward the day when whole-genome sequencing is standard of care. The system will ship with a handful of so-called "super panels," which are analytical workflows customized to interpret whole genomes within the context of specific diseases.
These super panels are intended to serve as "in silico gene tests" that will allow researchers — and, eventually, clinicians — to run any genetic test on any patient who has had their whole genome sequenced.
“The genetic testing lab of the future is a software platform where gene tests are apps," George Church, a professor of genetics at Harvard University and co-founder of Knome, said in a statement. "This will shift genetic testing from a fixed, lengthy process to a rapid and highly dynamic one that makes full use of the data contained in the entire genome.”
KnoSys comes with super panels for cancer, epilepsy, heart disorders, rare diseases, and other conditions, as well as a set of tools and libraries that will allow researchers to create their own super panels.
This approach allows researchers to look beyond genetic "hotspots" implicated in disease and explore regions upstream or downstream of these variants in order to gain a better understanding of disease mechanisms, Lee said.
In the case of cardiomyopathy, for example, a researcher might want to explore co-morbidities or the pharmacogenomic effects of drugs that are being considered for a particular patient. And while there are 47 core genes associated with the disease, there are "200-plus genes that are in the same networks that you — maybe for cost reasons — wouldn't look at normally, but because you have the whole genome, now you can do that."
The system "is designed to let the clinic develop those kinds of tests, package them up into an app, and run them with the click of a button," he said.
Such capability would be of interest to clinical labs, said Jakub Sram, director of client development at the City of Hope Clinical Molecular Diagnostic Laboratory, a participant in the early-access program for KnomeClinic. While Sram hasn't had access to a KnoSys 100, he said that if the cluster-based system offers the same analysis capabilities as the company's service offering, then it will be a good option for groups like his.
The key advantage of the Knome analysis pipeline, he said, is that it can rapidly filter tens of thousands of variants down to a handful — five or 10 — that are most likely to be related to a given disease. Doing this with in-house-developed pipelines, as City of Hope has done in the past, requires specialized bioinformatics.
Sram noted that he compared an analysis of 10 exomes performed by Knome with the same analysis run on his in-house pipeline and another run by researchers at the University of California, Los Angeles. The results were very similar. "The data was pretty good," he noted, adding that they would still like to do some "vigorous testing" before determining whether to bring a KnoSys 100 in house.
The system's price tag of $125,000 should be "very attractive for medical centers, universities, and academic institutions who don't have the infrastructure to do it on their own," he said, adding that the cost of the hardware and software together is "close" to the price of the hardware alone. Even groups that have some computational capability might prefer the plug-and-play system, he said, because it's easier than shopping around for hardware and then ensuring that the software runs on it properly. "It's better to have a system that has already been checked and approved and that you can buy from one vendor," he said.
And Sram agreed that clinical labs are hesitant to use cloud-based services for whole-genome analysis, preferring rather to retain those capabilities behind their firewalls.
"If you do diagnostics or clinical work, most hospitals and CLIA labs would not be comfortable having data on the cloud," he said. "They want to control who has access to the patient data" and IT administrators in such settings would view cloud-based options as "a risk."
While Sram stressed that he's still not decided whether to purchase the KnoSys, he said that there's a "huge unmet need" in the market for inexpensive analysis systems that not only offer genome annotation, but also provide information on the role of variants in disease. "The match between genotype and phenotype is the key," he said.
Currently, "there are databases and programs that can do certain parts, but [the researcher has] to put everything together. No one commercial program can do it all. I think that is the niche that Knome is trying to fill."