NEW YORK (GenomeWeb) – GenomeNext is preparing to formally launch a new product, called Olympus, next month that retains the underlying architecture of the company's original offering but has been expanded to include much richer analysis functionality, more automation, and support for large population studies, among other features.
GenomeNext launched in 2014 to commercialize software called Churchill, originally developed at Nationwide Children's Hospital in Columbus, Ohio, that offered tools for managing, analyzing, and interpreting large quantities of genomic information including whole-genome or exome or targeted panel data. Churchill still forms the core of the company's platform, but Olympus offers a much wider variety of features and capabilities beyond variant calling and analysis, according to CEO James Hirmas. "We've gone much further than that ... to handle the whole workflow from [laboratory information management system] all the way to the [electronic medical record]," he told GenomeWeb. Olympus represents a more "holistic approach" to bioinformatics analysis that GenomeNext and its clients believe is currently lacking in the market.
Specifically, GenomeNext's platform now includes preinstalled pipelines for germline, tumor-normal, and gene expression analysis including tools for annotating and interpreting genomic variants based on data from over 22 curated private and proprietary databases. Customers can now implement their own internally developed pipelines on the system, and the company has plans to add more pipelines at a later date. Olympus also includes tools for automatically uploading raw sequence files directly from sequencing instruments and moving the data through different processing, analysis, and reporting pipelines as well as better metrics for checking the quality of reads.
The revamped system also offers solutions to customer-reported challenges with tasks such as moving large quantities of data to and from the cloud-based solution, according to Hirmas. One specific issue that customers had was being able to visualize their datasets post base calling and sequence alignment. Many traditional visualization solutions require that bam files be local in order to run, meaning that users have to download the files first.
"If you have a 100 gigabyte Fastq file [as input to Olympus] your bam files are going to be about 100 gigs," he said. "If you are a [clinician] and you want to diagnose a patient and you need to look at the bam pileups for a variant, trying to download a [large] file with a not great network could take a while." Olympus addresses this issue with a preinstalled visualization tool that allows users to automatically load and visualize their data in the cloud. "We are completely away from people having to really handle or manage these large datasets or bring them down," Hirmas said.
Furthermore, the platform is able to support large-scale genome-wide association studies, according to Hirmas. For example, a customer could upload 400 tumor-normal samples to the platform as well as supporting clinical, demographic, and other data and perform pairwise comparison studies that assess over-representation of variants and genes. This particular capability, which will be formally unveiled next month, is already being used by two customers, Hirmas said, but he declined to disclose who these clients are.
As proof of its abilities in this arena, GenomeNext points to a test challenge it was involved in last year that was organized by Intel and Amazon Web Services. The company was tasked with processing a Phase III dataset from the 1,000 Genomes consortium that included whole genomes and exomes from 2,540 individuals across 26 different populations — more than 5,000 samples in total. According to numbers reported by the partners, the GenomeNext platform was able to analyze the dataset — some 77 terabytes of information — in about a week, processing about 1,000 samples per day. Details from the study will be published at a later date.
Its more comprehensive portfolio should make GenomeNext more competitive moving forward, Hirmas said. "The market is saturated with a lot of annotation and reporting companies, [but] there are not too many players actually commercializing secondary analysis pipelines," he told GenomeWeb. "The fact that we are upstream from all of those processes actually makes us very valuable and a great company to partner with."
Furthermore, its underlying algorithmic strategy — described in this Genome Biology paper — which ensures the same result even if samples are analyzed repeatedly, remains a strong selling point, particularly for clinical customers, he said.
"One of the things that a lot of groups don't realize is that their secondary pipelines are not reproducible and deterministic," which poses problems "when you start translating that into a clinical environment." Moreover, GenomeNext takes care of the necessary storage and compute infrastructure that customers need to analyze their samples at scale freeing them from the costs associated with installing and maintaining local infrastructure, which can add up quickly.
Also greater automation within Olympus results in a much shorter time to results, which should help sway new clients, according to Hirmas. According to the company's own estimates, it can analyze a 30-50x whole genome with an input file size range of between 85 and 110 gigabytes in under two-and-a-half hours. A 50-100x whole exome between 6 and 15GBs in size can be completed in under 90 minutes, while a 100x targeted panel ranging between 1 and 5GBs can be analyzed in under 30 minutes. In each case, those analysis times are irrespective of the number of samples, according to Hirmas.
In a case study listed on its site, the company claims that it was able to analyze 34 clinical exomes — more than 300 gigabytes of data — from raw reads processing through clinical results in under five hours. Ultimately, the goal is to be able to provide clinicians with results within a week at most of patients' first visit, Hirmas said.
So far, customer responses to the new GenomeNext platform have been positive, according to Hirmas, however, the company will continue to reshape and revise its platform as customer needs change. Earlier this year, it raised $1.2 million in seed financing from Hydra Capital and other private investors to further develop new features and tools for Olympus among other activities. "We are very much a plug and play solution," he said. "We can get a customer onboarded within weeks. It still takes some collaboration and communication on both ends ... [but] we are not looking at long implementation times."
In the last year, Olympus has gone through a number of validation studies with different customers and partners that demonstrate the value of the solution as well as highlight its comprehensiveness and accuracy. The details of some of these studies will be published later on, but Hirmas pointed to successes such as the most recent iteration of the Children's Leadership Award for the Reliable Interpretation and Transmission of Your genomic information (CLARITY) challenge as an example of what Olympus can offer.
Launched in 2012, CLARITY is a crowd-sourcing competition organized by the Manton Center for Orphan Disease Research at Boston Children's Hospital and Harvard Medical School's Department of Biomedical Informatics. Last year's challenge, dubbed the Undiagnosed challenge, involved 26 teams representing nine academic groups and 17 companies. Each team received genomic sequences from five patients with undiagnosed disorders and their families and were asked to submit reports suggesting candidate variants for the conditions in question. According to results of the challenge presented at the Boston Children's Hospital's Global Pediatric Summit last November, the report from Nationwide Children's Hospital took top honors, beating a total of 21 other entries — the runners up were teams from Invitae and WuXi Nextcode. Nationwide Children's team used the Churchill technology on which Olympus is built to analyze the datasets made available for the CLARITY challenge.
The revamped platform has also attracted customers, such as Sanford Health most recently, who have tapped Olympus to provide genomic analysis and data management capabilities for the Sanford Imagenetics initiative. Earlier this year, Sanford selected Translational Software's pharmacogenomics knowledgebase and PGx platform to generate reports for a bespoke pharmacogenetics panel that covers 42 different genes and assess patients' ability to metabolize more than 400 medications. More recently, Sanford formed the Sanford Children's Genomic Medicine Consortium, a network of seven children's hospitals that have agreed to combine their research capabilities, expertise, and resources to enable genetic and genomic discovery in pediatric healthcare. Hirmas told GenomeWeb that the health system will use Olympus for germline analysis initially but will eventually expand that to include tumor-normal and gene expression analysis.
GenomeNext offers customers a number of platform access options including a fully managed cloud solution, private cloud options, an option to run the company's secondary analysis pipelines locally, as well as a hybrid alternative that combines local hardware with cloud compute. Previously the firm charged $250 to analyze data from a targeted panel, $500 to analyze data from a whole exome, and $800 to analyze data from a whole genome including storage. Hirmas told GenomeWeb that the per-sample pricing model still remains, but that the specific costs depend on clients' requirements and needs.
Meanwhile, the underlying technology, Churchill, is available at no cost for academic research use. According to numbers from Nationwide, to date more than 400 academic research centers have used the software.