Computational biologists at Mass General are mining information from the genes and medical records of cancer patients to find clues to save more lives.

Massachusetts General Hospital Cancer Center researchers with advanced computational expertise are decoding the intricate genetics of tumors. In the process, they are revealing more about what drives the growth, spread and recurrence of cancer. “Computational biology, particularly cancer genomics, is revolutionizing our understanding and treatment of cancer,” says Gad Getz, PhD, director of bioinformatics at the Mass General Cancer Center and the Department of Pathology.

Massive amounts of information can be gleaned from a tiny sample of tumor tissue.

Cancer computational biology is the field of analyzing data coming from cancer samples in new ways. “The reason that we need to do this is because the amount of data being generated is much larger now,” explains Dr. Getz. He is a world leader in developing the genomic tools needed to gather vast amounts of biological information and analyze the cascade of genetic events leading to a cancer.

The Basis for Precision Medicine

“It’s the basis for precision medicine,” Dr. Getz says, referring to a major initiative at Mass General and nationally to tailor cancer treatments to genetic variations in a person’s tumor.

Gad Getz, PhD
Gad Getz, PhD

Therapies targeting the unique genetics of a person’s tumor in order to block its growth and progression have dramatically improved outcomes for some cancers. Better understanding of the genetic events will help oncologists predict the best course of treatment. It will help them predict what happens if they give a particular drug to a patient and when and if the cancer will come back.

Massive amounts of information can be gleaned from a tiny sample of tumor tissue. “We used to look at cells under a microscope and stain with dyes to find individual genes, but today we can sequence entire genomes of cells,” says Dr. Getz, who is also the Paul C. Zamecnik, MD, Chair in Oncology.

How Big is Big Data?

The genome holds our DNA — a complete blueprint of genetic instructions. Dr. Getz estimates that sequencing one person’s whole genome to find mutations is about 300 to 500 gigabytes of compressed data. “That’s on the order of 100 DVDs just for one patient’s genome,” he says.

The more samples of each type of cancer they have, the greater the statistical power to find the events and processes that cause cancer.

Multiply that by the many, many thousands of these files they have on individual patients’ tumors and the result is what people refer to as “big data.”

After sequencing a genome to find mutations, Dr. Getz and his colleagues then develop statistical algorithms, which are essentially instructions for how to solve a problem. The quest is to figure out which mutations are “drivers,” important to each type of cancer, and which are not. Drivers may be bull’s-eyes for targeting treatment.

Power of Numbers

There are so much data that Dr. Getz’s group at the Broad Institute of MIT and Harvard, where he has a joint appointment, had to move it to the “cloud.” There, files of any size can be securely stored offsite and accessed via the internet. They created a new platform called Firecloud and new algorithms for integrating and analyzing data. Firecloud is preloaded with data from the National Cancer Institute’s Cancer Genome Atlas — maps of the key genomic changes in 33 major cancer types.

Ephraim Hochberg, MD
Ephraim Hochberg, MD

Dr. Getz is also developing TumorPortal, an internet gateway for basic and translational researchers, clinicians and eventually patients. It will help them interactively explore and mine the data, as well as the results of analyses performed in FireCloud.

“TumorPortal will enable researchers from all over the world to add their tumor data to the cloud and jointly analyze it,” Dr. Getz says. The more samples of each type of cancer they have, the greater the statistical power to find the events and processes that cause cancer.

To advance this work, Dr. Getz is recruiting additional computational experts. He is also collaborating with other Mass General Cancer Center researchers, such as Ephraim Hochberg, MD, of the Hagler Center for Lymphoma. Dr. Hochberg and his colleagues are developing ways to tap the existing reservoir of information in medical records on the 200,000 cancer patients seen at Mass General over the past 60 years. They are using such methods as artificial intelligence to extract the information that would otherwise remain hidden to researchers.

These kinds of studies have already benefited Mass General patients with rare cancers, Dr. Hochberg says.

Changing Clinical Treatment

So, if Dr. Getz has genetic sequencing information on 50 patients with specific subset of brain tumors, Dr. Hochberg says, “we can help him put those findings into perspective with lab results, treatment and outcomes from the records of another 1,000 brain tumor patients.” The power offered by combining these two unique data sources should allow improvements in the ability to bring hypotheses generated by TumorPortal and FireCloud into the clinic.

These kinds of studies have already benefited Mass General patients with rare cancers, Dr. Hochberg says. For example, he and Jeffrey Barnes, MD, PhD, used computational methods to search all medical records to find all patients with a rare and extremely quickly growing cancer called Burkitt lymphoma. After analysis, they found that adding the drug rituximab to standard treatment improved survival and decreased the possibility of the cancer’s recurrence. That has changed clinical treatment of these patients.

Computational cancer researchers have only just begun to tap the field’s potential, Dr. Getz says: “It’s just the tip of the iceberg of what we can do.”

To learn more about how your support could accelerate these computational efforts, please contact us.