Skip to main content
Premium Trial:

Request an Annual Quote

Data Analytics Startup Medley Genomics Targets Genomic Heterogeneity; Initial Focus on Cancer


NEW YORK (GenomeWeb) – Startup Medley Genomics of Providence, Rhode Island, wants to harness its expertise in the analysis of genomic heterogeneity to help improve the treatment of cancer.

The Brown University spinout was founded last October by Ben Raphael, until recently a professor of computer science at Brown University; former Claritas Genomics President and CEO Patrice Milos; and Nabsys CEO Barrett Bready.

The company holds an exclusive license from Brown University to algorithms and software developed by Raphael, who is now a professor of computer science at Princeton University. Raphael holds a PhD in mathematics from the University of California, San Diego and has worked in cancer genomics for many years, including research for The Cancer Genome Atlas (TCGA) and the International Cancer Genome Consortium (ICGC).

Earlier this month, Medley Genomics, which currently leases office space in Providence from Nabsys, said that it has partnered with the Lifespan Cancer Institute to test its software on genomic data from primary and metastatic tumor samples. The project is supported with a $50,000 grant from the Rhode Island Commerce Corporation to Rhode Island Hospital, one of Lifespan's several locations.

Medley is currently financed with $250,000 in seed funding from the Slater Technology Fund, which has previously invested in Nabsys. Later this year, the firm hopes to raise an additional $900,000 to $1.5 million in angel funding to increase its staff, support early commercialization, and fund clinical utility studies. In parallel, Medley plans to apply for government grants through the Small Business Innovation Research (SBIR) program.

The company is built on the premise that improving precision medicine, in particular, personalized cancer treatment, requires a better understanding of the genomic heterogeneity of tumors, both within individual patients and between patients.

"Medley Genomics was formed to help us understand the molecular subtypes of a tumor," said Milos, who serves as the company's CEO, and the firm wants to commercialize software tools to analyze tumor sequencing data. Much like with HIV, the idea is to find multiple vulnerabilities in different subpopulations of a tumor and attack them with several targeted therapies in combination in order to "make cancer much more of a chronic disease," she said.

"What we really need to start thinking about is treating tumors more as an ecosystem, an evolving system," Raphael said. "What we want to get to within Medley is to transform some of the precision medicine programs at many of the large cancer centers to not being focused on 'let's do a precision medicine approach on a tumor,' but 'let's do a precision medicine approach for the population within that tumor.'"

He cautioned, though, that there will be challenges – genomic analysis might find two subclones within a tumor, for example, that could be treated with two different targeted therapies, but the two drugs might be toxic in combination.

During his time at Brown, Raphael and his colleagues developed a number of algorithms that can be used to analyze intra- and inter-tumor heterogeneity from large-scale sequencing datasets, and Medley plans to develop these tools further for commercial applications.

In late 2014, for example, his group published a study in Nature Genetics, in which they used an algorithm called HotNet2 to analyze mutated subnetworks of genes in almost 4,000 TCGA samples from 12 cancer types and found both well-known cancer signaling pathways and subnetworks with less well-defined roles in cancer. Overall, the HotNet algorithm has been used in about a third of all TCGA publications, Raphael said. Being a fairly general tool, HotNet could also be used to describe heterogeneity based on input from other data types, such as RNA or metabolite data, Milos added.

In another area, Raphael and his colleagues have developed algorithms that look for mutually exclusive mutations between tumors, including single nucleotide variants, copy number variants, and structural variants, and use those results to find pathways that are perturbed. Those tools – such as Dendrix, Multi-Dendrix, and WExT – have been used successfully in TCGA cohorts with different cancer types, he said.

Raphael's team has developed other tools, for example, THetA, AncesTree, and SPRUCE, to tease out mutations present in subpopulations within a patient's tumor, either from bulk tumor sequencing data or from data from several sections of a tumor. Some of the tools go a step further and define evolutionary relationships between those subclones, building phylogenetic trees. "This is critically important for treatment, as well, because if we just treat a tumor using the precision treatment approaches available, we may only be treating the dominant subclone, and there might be subclones of tumor cells in minor proportions that would not be responsive to this targeted therapeutic," he said.

Subclones defined by using these tools in the primary tumor could also be subsequently monitored in circulating tumor DNA. "Through tracking of the patient with later liquid biopsies, we could get a sense of which of the subclones might be circulating at different frequencies," Raphael explained. "Although we haven't done that from any data from liquid biopsy yet, we have done similar things with targeted sequencing data from the primary tumor."

All of the tools developed by Raphael's group have been freely available to academic researchers, and they are encouraged to continue using them, he said. What Medley plans to do is bundle the tools into a "scalable, robust solution" for analysis of intra- and inter-tumor heterogeneity by commercial players, such as pharmaceutical companies or diagnostic laboratories. The idea is to put the tools together "in a way that goes from genomic data to more of an answer," he said, which his academic group was not equipped to do.

Milos said that the software is already very robust, and that Medley plans to make it more scalable, creating cloud-based interfaces that will enable users to have either their own domain with the tools on the cloud or to bring the tools in house. Combinations of tools could also be customized, depending on a user's needs.

Up until now, Medley's tools have predominantly been used for discovery research projects, but in its collaboration with Lifespan, the company will start to explore their clinical utility for patient treatment.

For the retrospective study, which is about to start, Medley will analyze 50 or more tumor samples from Lifespan's collection, including primary and metastatic tissue samples. Sequencing for the study will be outsourced to a service provider. The idea is to look at a patient's therapeutic response and "to begin to build a correlation between the clinical prediction from the primary and what happens in the metastatic tumor," Milos explained. The study will also involve methods validation, she added.

Medley is already in early-stage talks with several pharmaceutical companies about adopting its tools for translational research, as well as with informatics firms that focus on tumor analysis who might be interested in adding the algorithms to their offering. The company has also had conversations with academic medical centers about placing some of Medley's software into their existing pipelines as part of research collaborations.

"We're interested in multiple avenues," Milos said. "The key thing we think is important as a company is to be able to support these tools and be knowledge experts for them."

Longer term, Medley might develop its tools for use in other areas besides oncology. "Clearly, genomic heterogeneity doesn't just exist in cancer," Milos said. For example, it occurs in the prenatal setting, where circulating nucleic acids derive from both the mother and the fetus; in the microbiome; and in immunology. "The idea is to continue to look for novel approaches to genomic heterogeneity, but right now, we have a nice focus in cancer," she said.

Medley's competitors include academic researchers, among them those studying tumor heterogeneity through single-cell analysis. Also, clinical cancer profiling and analysis firms, such as Foundation Medicine and Tempus, might start paying more attention to tumor heterogeneity in the future, Raphael said. In addition, companies like Grail that specialize in circulating tumor DNA testing might become competitors going forward, once they focus more on the analytical side.

"It's certainly a competitive landscape but also a very complex scientific and analytical challenge," Raphael said. "We imagine that there are potentially some interesting partnership opportunities with some of the companies that are doing the sequencing or have novel technology approaches but may not have the analytical approaches to maximize the utility of their data," he said. Medley is already talking to a few of these firms, he added.