CUSTOMIZING TREATMENTS BASED ON INDIVIDUAL’S GENOME

SHILO REA

An oncologist delivers a cancer diagnosis to her patient, who is visibly and understandably upset. As the doctor begins to outline treatment options, the concerned patient listens intently, holding on to hope that the plan that the doctor recommends will work for her. The doctor bases her recommendation on the fact that the treatment has shown success in people with similar types of cancer. Weeks later, after several grueling rounds of treatment, the doctor informs the patient that the cancer has not responded to the drug and offers a new drug therapy to try. Precious time has been lost and the patient’s fear for their survival heightens.

It is a story that is far too common.

School of Computer Science researchers are working to remove some of the guesswork in cancer treatment. Ocean Genomics, a Carnegie Mellon University spinout co-founded by Carl Kingsford, envisions a future for medicine where an individual’s genetic makeup informs these decisions. By providing drug manufacturers and physicians with better tools, Ocean Genomics can assist them to develop and prescribe specific treatments for individual patients.

Kingsford, the Herbert A. Simon Professor of Computer Science, has spent his career creating novel computational methods that analyze genomic data. Specifically, he has focused on ways to interpret what happens when a person’s DNA decodes instructions for making proteins or other molecules — a process called gene expression.

Kingsford’s recent work involves designing techniques to analyze the sequences of messenger RNA (mRNA) molecules, which are the products of gene expression and indications of each gene’s activities. “Drug companies and researchers have the data. We developed a number of tools to analyze gene expression data and they became popular,” Kingsford said. “It’s not easy to support [the tools and researchers] in an academic context, so we started Ocean Genomics to build and extend these tools to be accessible and useable in translational contexts.”

Combining machine learning and artificial intelligence techniques, Ocean Genomics provides a cloud platform for users to perform gene expression analyses and access to expert support services. This allows users to compare the RNA expression levels of multiple genes in samples under various conditions and can give clues into how gene expression differs between healthy and diseased samples, and in response to therapies.

Ocean Genomics’ tools offer high computational efficiency with speed and accuracy, a significant deficiency for other available programs. For example, once 50 samples are uploaded, results are typically available in a single day.

The company’s early success points to it leading a transformation in the field of precision medicine. They recently partnered with researchers from Samsung Medical Center in Seoul, South Korea, to predict which gastric cancer patients will not respond well to the popular drug Pembrolizumab — all based on their gene expression features.

Pembrolizumab, an immunotherapy drug that targets and blocks PD-1 proteins in T-cells, prompts the T-cells to attack cancer cells. Some patients respond very well to the treatment while others develop resistance after the initial round.

Ocean Genomics’ platform extracted features, generated classifiers, performed quality checks and delivered comprehensive analyses from 55 patient samples and related metadata within 24 hours. They provided early evidence suggesting seven genes that may be related to Pembrolizumab becoming ineffective in those patients, providing the basis for potential follow-up studies.

“This type of insight will help providers determine which treatment will help which patients based on biomarkers,” said Kingsford. “And it will help treatment makers develop more targeted options.”

In order to move this work further, Ocean Genomics is also involved in working with drug companies on biomarker development. “We are translating ideas from the DNA world into RNA-based biomarkers. Immunotherapies and cancer therapies are expensive to create and to administer. Since they don’t always work for all patients, we are looking at how they could be improved,” Kingsford said.

In another promising discovery from a large clinical trial in Japan, Ocean Genomics identified an early signal for a gene expression pattern that may be predictive of whether a patient will respond to a particular drug.

Outside of cancer treatments, the team currently works with a pharmaceutical company on a neurodegenerative disease with their mRNA analysis tool for single cells. Dr. Stanley Marks, chairman, director of clinical services and chief medical officer of UPMC Hillman Cancer Center, who is on Ocean Genomics’ scientific advisory board, said, “RNA identification can guide therapies for patients, and Ocean Genomics is actively engaged in moving the field.”

In March, the company announced a strategic partnership with Geninus, a leading full-service genomics laboratory out of Seoul, South Korea, to advance research and clinical applications with hospital, academic and biotechnology clients. The companies will launch Ocean Genomics’ advanced AI-based transcriptome analysis and biomarker platform (txome.ai) on the cloud and on location in Korea. Geninus will use txome.ai to expand its CancerSCAN clinical diagnostics program, and the two companies will also co-develop a series of RNA-informed predictors of drug response for research and clinical use.

“Genome analysis for precision cancer medicine is becoming an essential process in hospitals,” said Woong Yang Park, Geninus CEO. “Gene expression analysis on tumor tissue RNA can deliver critical information for targeted therapy and immunotherapy.”

Through Ocean Genomics, Kingsford reinforces his personal commitment to mission-driven work by expanding the reach of accurate gene expression analysis. In late 2020, the company received a $235,000 grant from the Chan Zuckerberg Initiative, a new kind of philanthropy dedicated to using technology to help solve some of the world’s toughest challenges, including eradicating disease.

While it’s unusual for a business to receive CZI funding, the award will allow Ocean Genomics to improve their open-source projects Salmon (bulk RNA sequencing) and Alevin (single-cell RNA sequencing). It will also fund creating tutorials and documentation so more researchers and practitioners can use their advanced features and contribute to the projects.

Increased participation could lead to more rapidly developed personalized cancer treatments, giving patients the benefits of science and data-based treatment options, as well as the peace of mind they need to battle this deadly disease. ■