SPOTLIGHT ON CANCER RESEARCH

Cancer genomics: Collaborative research

As sequencing costs reduce and computing power expands, opportunities abound for scientists to learn about the genetics of cancer.

WHEN ROBERT Strausberg started his career in 1976 as a microbiologist, studying yeast as a postdoc at Southwestern Medical School, he had no idea that one day he'd be working in cancer genomics.

Credit: ISTOCK PHOTO/THINKSTOCK

From a field that did not exist, it has become a promising new direction for understanding and treating cancer, attracting researchers from across scientific disciplines, some of whose expertise might not have seemed immediately useful for cancer research. “I didn't think what I was doing applied to much else,” says Strausberg, now executive director of collaborative sciences at Ludwig Cancer Research in New York. But it turned out his experience was more useful than he thought. An early-stage biotechnology company, Genex, hired him to apply his knowledge of yeast to develop vaccines and drugs. His studies of yeast's mitochondrial DNA and what he'd learned about the physical mapping of DNA eventually won him the head role at the Sequencing Technology Branch at the US National Center for Human Genome Research among various other jobs at the National Institutes of Health. Because genomics is basic to biology, Strausberg's increasing knowledge helps him understand various fields important to human health, from infectious disease to agriculture. “My interest in genomics has opened a lot of doors for me.”

Genomics is the study of the set of genes within an organism. As gene sequencing costs decrease and computing power continues to grow, it's become possible to collect and analyze the information contained within a genome. In cancer, that translates to figuring out which mutations are involved — not only by cancer type but by individual patient. Researchers are learning about the causes and progression of tumour evolution, identifying targets for therapy, and seeing which genes may make a particular drug more or less likely to be effective.

Genomics is increasingly being applied to the clinical treatment of cancer, as well as to research into cancer biology. It also contributes to targeted cancer therapy, where treatments are aimed at specific genes or mutations in a particular patient. This approach can prove more effective and less toxic than traditional chemotherapy, which kills cells indiscriminately. “Cancer genomics for me is an opportunity to develop new treatments for advanced cancers, to learn how the disease develops in the first place, to do better detection and prevention,” Strausberg says. “To me having a specialty in genomics gives one a very broad field of opportunity.”

There's plenty of funding for the field too. In the US, the National Cancer Institute has created the Cancer Genome Atlas, which is collecting genomic information about more than 20 different types of cancer and making them available to researchers for analysis. The US National Human Genome Research Institute collects genetic research on several types of cancer and provides funding for university scientists. Cancer Research UK has created a multidisciplinary project award to encourage research by multi-discipline teams and plans to fund about 10 projects with up to £500,000 over four years.

Varied expertise

Cancer genomics exists at the intersection of clinical work, biological research and computer science. Therefore, there's a need for researchers from various disciplines — medicine, molecular biology, chemistry, computational modeling, bioinformatics, mathematics, even engineering and physics. Whatever their speciality, scientists entering the field need to be conversant with other disciplines and be comfortable working with researchers from different areas. “You actually need specialists from different fields working together to understand what is going on,” says Moritz Kircher, who is both a clinical radiologist and a researcher in molecular imaging at Memorial Sloan Kettering Cancer Center in New York. “No-one can alone really tackle those questions anymore.”

Kircher works in radiogenomics, attempting to correlate information from imaging techniques such as magnetic resonance imaging with sequenced genes. If specific genes have specific effects on the image of a tumour, doctors might get a better idea of what's happening with a cancer overall, and progressively, than they would from taking a biopsy from only a small area of the tumour.

Kircher's also developing nanoparticles that use the signatures of genetic mutations to home in on tumour cells and let pathologists image more detailed information. His lab, for instance, recently developed Raman nanostars, tiny, star-shaped gold particles coated in silica that shine brightly under laser light. Early tests on mice showed the nanostars could detect tumours as small as 100 microns, as well as premalignant lesions.

In the UK there's a shortage of people able to maintain the datasets and analyze the sequencing data being produced, says Ultan McDermott, a team leader in the Cancer Genome Project at the Wellcome Trust Sanger Institute in Hinxton. A lot of the initial work that used sequencing data from a few dozen or a few hundred samples to create and validate algorithms is done. The challenge now, McDermott says, is to scale up the algorithms so they can work with hundreds of thousands of samples, and to make sure people from outside the group that developed the programs can make use of them. “Developing these is going to be a lot of work and hugely important,” McDermott says.

Much depth, some breadth

It's a good idea, researchers say, for people from one scientific discipline to have some understanding of others they'll be dealing with in genomics research. Biologists entering the field should have some grounding in statistical analysis and some sense of how computational modeling works in order to make sense of the quantitative data they're looking at. “I wouldn't suggest medical doctors or molecular biologists start doing a whole course on computational models or bioinformatics,” says Andrea Sottoriva, head of the Evolutionary Genomics and Modelling Team at London's Institute of Cancer Research, himself a computer scientist with a physics background (see From neutrinos to tumours). “But I would suggest definitely to take a course in statistics.”

Computer scientists, on the other hand, need at least a basic understanding of biology. “It's essential, so they understand they're not just a bunch of numbers, that there's a person as the other end of these measurements,” says Nick Luscombe, a computational biologist who studies gene expression at Cancer Research UK's London Research Institute and University College London. He also runs a smaller laboratory at the Okinawa Institute of Science and Technology in Japan. McDermott says that such people, sometimes come from physics and engineering backgrounds, are quite difficult to recruit. “These individuals don't always appreciate there's a lot of interesting research they can do in the cancer field.”

There are also not many cancer fellowships and funding calls geared toward computing and mathematics. But the work of data scientists could also increase opportunities for biologists, he predicts, by democratizing the use of genomics. Biologists who aren't part of large cancer research centers will have access to these data sets and algorithms through cloud computing, and be able to make their own contributions to fighting cancer. McDermott says that some people believe future cancer research groups will be a 50-50 mix of data experts and wet lab scientists.

Another option is to do online courses like the Genetics and Genomics Certificate offered by Stanford University. The course requires two core courses in the fundamentals of genetics and genomics, plus four elective courses on topics such as the genetics of cancer or the applications of gene therapy. It can be particularly useful for doctors treating cancer patients, says Michael Snyder, director of the Center for Genomics and Personalized Medicine at Stanford. “Many of these people got their MDs 15 years ago and the field didn't exist,” he says.

Having a broader knowledge doesn't mean someone should aim to be an expert both in biology and computation, Luscombe says. “You run the risk of not being good enough at either,” he warns. Kircher agrees: “in general you want to have one area where you are really, really strong and maybe one or two other areas that are really complementary to that,” he says. One way to broaden expertise, he suggests, is to do more than one post-doctoral fellowship in different labs.

Physicists or mathematicians might be wary of moving into biology, which may seem absolutely alien to them. But the novelty of cancer genomics means they may be able to make bigger contribution than they could in more mature field, Kircher argues, and without a multi-year, multi-billion-dollar international effort. “There is so much stuff that has not yet been done,” Sottoriva says. “You don't have to wait for another Large Hadron Collider to be built to answer questions in this field.”

From neutrinos to tumours

Credit: ANDREA SOTTORIVA

Many cancer researchers started out as undergraduate biology students, but not Andrea Sottoriva. Now head of the Evolutionary Genomics and Modelling Team at London's Institute of Cancer Research, Sottoriva spent much of his time as an undergraduate and graduate student as a programmer for an experiment trying to detect neutrinos, elusive subatomic particles that can pass through the bulk of the Earth without slowing down.

But while he was working on his master's degree in computational modeling at the University of Amsterdam, which he finished in 2008, he started thinking about a different area of research. “There was this emerging field in bioinformatics and it caught my attention,” he says. He was drawn by the opportunity to apply his research directly to health care, as well as by the recognition of how many open questions there were in this new field that computational modelling might address. He went on to earn a Ph.D. in cancer genomics and modelling from Cancer Research UK's Cambridge Research Institute, then did a post-doc at the Norris Comprehensive Cancer Center at University of Southern California, in Los Angeles, where he studied how genomic sequencing of multiple samples from a cancer patient could help in the understanding of how tumours evolve.

Cancer research is no longer solely the domain of biologists and medical doctors, thanks to the burgeoning field of genomics. “The cool thing is now in biology we have all this quantitative data, this digital data that we need computer technicians to analyze,” Sottoriva says. “It is possible to really jump into this field which is becoming more and more interdisciplinary.”