NEW YORK (GenomeWeb) – Bioinformatics firm Genestack is expanding its business to include the clinical genomics market and has revealed the UK's Cancer Molecular Diagnostic Laboratory (CMDL) as its first customer in the space.
The company is eyeing the translational genomics market where clients are working on moving resources developed within the research context into clinical applications, Genestack CEO Misha Kapushesky told GenomeWeb. As the company tries to push its bioinformatics tools into clinical use, its pitch is that the platform provides the "foundation" for clinical researchers to build on.
"It's a universal platform [that] combines ... end-to-end data processing from raw [sequence] to report with the ability to conduct research and dig deeper into the data," he said. "It fits especially well into the translational context, where one wants to go rapidly from research to application."
In addition to offering cloud- and cluster-based access to its solution, Genestack has developed an appliance in partnership with Intel that couples its platform with Intel-developed processers. It will implement the first of these systems at CMDL, a collaborative venture that involves researchers from Cancer Research UK, the University of Cambridge, the Cambridge University Hospitals NHS Trust, and the Medical Research Council.
The lab has tapped Genestack's platform to analyze data for prognosis and post-treatment monitoring of bone marrow transplant patients, with an eye towards improving patient care as well as saving time and money. Specifically, it will use an appliance to analyze data from a chimerism assay it has developed, which assesses the extent to which donor grafts have been accepted in bone marrow transplant cases. It is one of several tests that CMDL is developing for eventual use in the clinic that could leverage the Genestack appliance, George Vassiliou, lead clinician for CMDL-Haematology, told GenomeWeb.
The chimerism test, which CMDL hopes to roll out in the next few months, evaluates how much of the patient's blood is made by the graft cells and to what extent the patient's own cells are making blood. If a greater proportion of the patient's own cells are still making blood it could mean that the transplant failed, or even that the cancer is returning.
"At the early stages of the transplant in particular, and also in the late stages, you keep a close eye on what percentage of the blood is [made] by the donor and the recipient," Vassiliou explained. "This [test] is a very efficient and quick way for determining that. It uses the fact that the genes between the donor and the recipient differ, and it can tell by using DNA what percentage is donor and what percentage is recipient."
CMDL's partnership with Genestack developed through a funding scheme from the UK's Medical Research Council, Vassiliou said, that was intended to help the lab find an informatics partner for its tests. CMDL develops diagnostic solutions for cancer that it plans to push into clinics, and it needed an easy-to-use informatics solution that researchers in hospital labs could understand and use, he said. It also wanted a system that wasn't already tied to an existing test or platform and was general enough to support multiple tests. In addition to meeting those requirements, Vassiliou said he was also impressed by Genestack's pedigree and expertise. "[They] know what they are talking about and they are genuinely interested in developing something in the clinical space," he said. "It's a good fit."
Founded nearly five years ago, Genestack is rooted in projects that Kapushesky was involved in at the European Bioinformatics Institute. At EBI, he led a functional genomics research team that was responsible for developing the Expression Atlas, a database of gene expression patterns that have been observed under different biological conditions.
Kapushesky's team developed Expression Atlas to help the community better access large datasets at the EBI, he told GenomeWeb. "We were sitting on a lot of public data, which was pretty much held in static zip files, and we wanted to make it queryable," he explained. "We wanted people to [be able to] type in the name of a gene, protein domain, a pathway, tissue, cell line, or disease condition and know what happens where." In total, the researchers provided public access to data from some 140,000 different assays across roughly 7,000 different experiments and 6,000 cell lines, he said.
The resource received a lot of interest from researchers in both pharma and academia and continues to be used in gene expression research today. It is also one of the data sources for the Centre for Therapeutic Target Validation platform, which was launched by the EBI, GlaxoSmithKline, and the Wellcome Trust Sanger Institute. The platform gathers and integrates information for target identification and prioritization.
But clients soon began asking the Expression Atlas developers for tools that would let them combine the publicly available data in the database with their own private data, according to Kapushesky. They also asked for more visualization tools and the ability to run different kinds of statistical analyses on the combined data than was possible at the time.
That's when "we realized that we needed to build a platform that will track how data gets in [and] what transformations happen to it [as well as] allow you to configure what tools are used, plug in whatever statistics you want, define whatever visualizations you want, and combine your data with data from collaborators and from the public domain," he said.
Kapushesky and some EBI colleagues would eventually spin off Genestack from the institute to focus on commercializing the platform. Since its launch, the company has primarily pursued opportunities in the life science research space. For example, it partnered with Rothamsted Research to develop agrigenomics technologies for crop improvement in 2015 — the partners received $138,000 from Innovate UK for the project.
In 2012, the company partnered with Constellation Technologies on one of three pilot projects for the Pistoia Alliance, a non-profit consortium of life science companies, vendors, publishers, and academic groups. The pilots were organized by the consortium's sequence services group as part of efforts to build a platform for NGS data storage and analysis for pharmaceutical research and development use. Genestack has also worked with some clients outside the space, such as consumer goods companies, according to Kapushesky.
However, "we found that a lot of research needs translate into the clinic well, [such as] a focus on reproducibility and data governance, understandable reporting, robust pipelines, careful attention to metadata, and data security," he told GenomeWeb. These are the sort of needs that Genestack is designed to address, he said.
Attention to metadata is one of the features that Kapushesky believes sets Genestack's system apart from other analysis platforms in the market that companies like DNAnexus offer. Genestack's platform uses controlled vocabularies and ontologies to capture and describe internally hosted and third party datasets. It uses a common set of metadata descriptors that make it easy for researchers to combine and query disparate datasets and "slice and dice" them in different ways.
The system also offers templates for describing new datasets in the same language and terms used to describe existing data. Users can build interactive applications within the platform that let them visualize sequences and variants in different ways, and they can add these visual representations of data to their reports.
Genestack can take in whole genomes and exomes, and methylation data, among other kinds of data, and will soon be able to accept proteomic information as well, Kapushesky said. Currently, it provides access to information from public repositories, such as ArrayExpress, the Gene Expression Omnibus, the Short Read Archive, and more.
Genestack charges an annual license fee for its software, whether customers purchase it as an appliance or on the cloud. In addition, there are compute and storage costs associated with running the platform in the cloud. The company is not disclosing pricing details for either the cloud or the appliance options at this time. In addition, Genestack will offer different appliance configurations, and customers will be able to combine multiple boxes into a cluster easily. There will also be an option to deploy a hybrid system, where customers can run some analyses on their appliance and others on the cloud, Kapushesky said.
The platform is compatible with all the major cloud infrastructure providers, including Amazon and Google. Customers can also deploy the system on internal clusters if they don't want to purchase the appliance or use it in the cloud. This flexibility appealed to CMDL, Vassilou noted. The lab is implementing Genestack as an appliance, but it is valuable to have a way to connect and collaborate with other hospitals and clinical centers if the researchers want to.
Genestack will continue tailoring its platform to the clinical genomics market. It has partnerships with other clinical researchers that it will disclose later this year, one of which focuses on a test for identifying and reporting infections in cystic fibrosis patients. CMDL, meanwhile, is already looking ahead to other potential use cases for the Genestack platform. Vassiliou told GenomeWeb that the lab is looking into using the appliance for immunotherapy and minimal residual disease tracking assays, for example.
"We are at the very start of this process and there remain a lot of unknowns," Kapushesky said. For example, "it is clear already that we have to work a lot on making our reports and visualizations even simpler and more accessible to medical professionals whose time is limited," he said. Also, "we have to separate out research-related aspects and reporting-related aspects more strongly."
The company also hopes to see the appliance adopted more broadly in the non-clinical market, Kapushesky said, with current applications developed for CMDL and other clients serving as blueprints for designing appliances in those domains.