Like many people in the early 1990s, Lynn Bry -- then a student in the M.D.-Ph.D. program at Washington University in St. Louis -- didn't know much about computers. A computer scientist friend in Pennsylvania had asked Bry to write some content for a Web site he'd built. "I thought I could just put what he wanted on a disk and send it to him, but he said no, use FTP," she says. "And I was like, FTD? What do flowers have to do with computers?"
Lynn Bry "worked through a ton of regulatory issues that needed to be thought through. A lot of people have great ideas, but they don't stick with them. She stuck with it, and there's a lot to be said about that. To me, she symbolizes persistence." --Shawn Murphy
Then her Pennsylvania friend got the administrator in Bry's department to set her up with an e-mail account. "That was another bizarre conversation," she remembers. The administrator "was talking to me about UNIX, but I heard 'eunuchs.' "
Bry persevered, learned how to set up a web server, the basics of UNIX administration, and how to script in Perl. Later, she taught herself Structured Query Language (SQL), a programming language used in database management.
Now an assistant professor of pathology at Harvard Medical School, Bry, 42, has melded her computer knowledge with her clinical and research education to fill a critical need: She has developed an informatics solution to get blood and other biological samples to researchers at a lower cost, and in a shorter time frame, than ever before. Through her efforts developing a computer application called Crimson, Bry is hoping to accelerate the pace of translational research.
This is part of an article series for CTSciNet, the Clinical and Translational Science Network, an online community. These articles are published on both Science Careers and within CTSciNet. If you would like to see your organization's articles on clinical/translational science careers included in our growing comprehensive resource, please send an email to jaustin(at)aaas.org.
Bry "has developed a highly innovative and nationally recognized system to use the biological samples obtained routinely in the course of clinical care as the basis of population-based discovery research," Isaac "Zak" Kohane, director of the Children's Hospital Informatics Program and professor of pediatrics and health sciences and technology at Harvard Medical School, writes in an e-mail. "Dr. Bry has used her clinical training to inform herself of scientific priorities in both her basic and applied research. From my perspective, this makes a world of difference in identifying those challenges that are central to translational research."
Learning the Web
Bry's interest in biology began as a child on Hilton Head Island in South Carolina where she spent a lot of time catching seafood and studying the strange organisms that washed up on the beach. Then, as an undergraduate at Cornell University, she took a microbiology course and "a whole world opened up. I used to tell my roommate that I felt like I needed a seatbelt in microbiology class."
On her way to a bachelor's degree in biology, Bry was considering graduate school to study plant genetics and soil microbes. But after she completed a summer stint on the genetics of Escherichia coli at the Scripps Research Institute in San Diego, California, Bry's mentor there asked if she had an interest in infectious diseases or anything medical and suggested that, if she did, an M.D.-Ph.D. program might be a good fit.
Bry ended up in the M.D.-Ph.D. program at Washington University in St. Louis, studying with Jeffrey I. Gordon, now director of the university's Center for Genome Sciences and Systems Biology. "Rather than soil microbes and plants, I studied the bugs in your gut and how they influence it," Bry says. While conducting research for her Ph.D., Bry began to learn about Web sites, servers, and programming.
Around the same time, she became involved with a program in which university students and researchers visited local public schools to give presentations. "Invariably, the kids had questions after we left, but we had no avenue for them," she says. Bry and a few friends realized that the Internet offered a solution. "We spent a couple of late nights writing code and got people to volunteer to answer questions."
The result was the MAD Scientist (MadSci) Network, an ask-the-scientist site that was getting nearly 2 million hits a month at a time when the information superhighway was still a dirt road. The network now has an index of more than 36,000 questions, answered by experts around the world (including Bry), addressing questions such as why grapes spark in the microwave and why songs gets stuck in your head.
The site won several awards and wide acclaim. "[The MAD Scientist Network] was a nice example of how you could leverage the power of the Internet to bring people together, in the days before Facebook and whatnot," Bry says. "That's where I learned a lot and how to do things in high-throughput fashion." Later, Bry was invited to help the U.S. government create a similar site, now defunct, which was used by libraries.
Linking research and samples
After graduating in 1998, Bry went to Boston for a 3-year pathology residency, followed by a postdoc in molecular immunology in the lab of Michael B. Brenner, now chief of rheumatology, immunology, and allergy at Brigham and Women's Hospital. She stayed in Boston to take her current position at Harvard Medical School, where today she maintains a lab for her research on mucosal immunology and host defenses against microbial pathogens. She spends about half of her time on Crimson and other biorepository projects; she is also the director of the Brigham and Women's Specimen Bank and the Partners Biorepository for Medical Discovery.
The idea for the informatics infrastructure that would become Crimson came around 2002, while she was doing her postdoc. Colleagues in the rheumatology department viewed Bry, the lone pathologist, as a potential source for biological samples. They would ask her, for example, for 100 blood samples from people with rheumatoid arthritis to test for possible diagnostic markers. Bry figured she could write a computer program that would comb the biorepository databanks and find acceptable samples.
The program worked, but Bry soon realized that she needed something more sophisticated -- an application that took into account regulations and privacy checks, kept records of requests, and allowed for release of data and samples, all in a timely fashion. She approached Neil Herring, a consultant for Partners HealthCare, a nonprofit health system that includes the Harvard-affiliated hospitals. "If you needed something to work, you talked to him," she says. "He's been 50% of the brain trust of Crimson."
The two met in a coffee shop, scribbling ideas on a stack of napkins. They worked up a high-level specification for what they needed and spent almost 2 years seeking out funding. Michael Gimbrone Jr., chair of the hospital's pathology department, approved the idea and gave them pilot funding. Three years of work with Cambridge, Massachusetts–based Daedalus Software produced Crimson, which launched in 2007.
Crimson receives daily feeds of information on biosamples -- mostly blood, but also urine, tissue, cerebrospinal fluid, and bacterial collections -- from patients in Harvard-affiliated clinical laboratories and the pathology laboratory, which includes some 5000 blood samples collected daily at Brigham and Women's and Massachusetts General hospitals. "Anything coming through a clinical lab, we have data on it," Bry says.
Crimson matches samples to known research needs. The matches are run through an "honest broker" engine that determines which data come with each sample, ensuring privacy and compliance. "Most of our samples are consented -- that is, the patient is in a clinical trial or has consented to have the sample used in a clinical trial," Bry says. Samples also can be de-identified, allowing them to be used without consent in situations where patient identification is not necessary.
Crimson keeps track of what's been requested and released, and how quickly each research study is accruing samples. It includes an inventory-management system, including bar-code printing and shipping functions. It also has a fee structure and billing system; research groups pay to access samples through Crimson and for specimen shipping, collection, and storage. This helps the application to support itself financially.
"When you create an information systems solution, you have to look at it as an end-to-end process," says Herring, the consultant who designed Crimson with Bry. "You can't just implement something that does the sexy piece; it has to allow you to carry information about investigators, it has to allow you to send them a bill. What good is it if you can't implement a business process? Lynn got that right from the get-go."
Bry also spent a lot of time considering regulatory oversight, privacy, and compliance issues. All of these factors added complexity to the research tool. "In the end, it was persistence in solving a problem" that led to Crimson's success, says Shawn Murphy, assistant professor of neurology at Harvard Medical School and medical director of research computing for Partners HealthCare. "She worked through a ton of regulatory issues that needed to be thought through, and there was a lot of pushback from the [institutional review board]. A lot of people have great ideas, but they don't stick with them. She stuck with it, and there's a lot to be said about that. To me, she symbolizes persistence."
Enabling translational research
Crimson handles 40 to 60 research studies each year, and since its launch has provided more than 100,000 samples to researchers and helped support more than $30 million in research grants.
Among those are grants for very high-throughput genomics and proteomics studies. One research group needed 20,000 samples for a study of the epidemiology of common diseases. Using standard clinical trial channels, about 50 samples per month could be collected at a cost of $1200 per sample. That's a $24 million study that requires 67 years for data collection. But with Crimson's informatics power, accruing the 20,000 samples will cost $170,000 and take less than 5 years. "The fact that you can rapidly accrue samples greatly reduces the costs and barriers," Bry says. "We can drop the cost by two orders of magnitude and increase accrual rates by one or two orders of magnitude."
Given the amount of data being generated daily, and the continued explosive growth in technology, it's no surprise that Bry sees bioinformatics as a growing area with "definite, immediate" needs for computer scientists, engineers, informatics experts, and statisticians.
"We have clinical data, genetic data, proteomic data. ... Increasingly, we will need to merge those data sets and either do fishing expeditions for what's important or develop ways to validate the information and find ways to use it," she says. "We'll need bioinformatics tools; we'll need statisticians; we'll need people who can build computational tools. And that's just on the research side. Layer on top of that the need to do these things for clinical decision support -- that's basically an empty field."
Getting Into Bioinformatics
Despite a lack of formal training in computer science or informatics, Lynn Bry has a breadth of computer-related knowledge. That enables her to apply information technology to achieve specific research goals, she says. "I know enough about software to know how things need to be structured … and then how do we operationalize it," she says. "It's also important to understand the economics; these activities are not cheap."
Bry believes a degree in bioinformatics is a good start, but "if you want to be managing projects and people, as well as managing budgets and doing strategic planning, an M.B.A. can give you those skills," she says.
Bioinformatics is a broad term that applies to creating and using computer applications and information sciences in biomedical research and health care. Focus areas range from the molecular (genomics, proteomics) to individual (clinical research, health systems management) to populations (public health) and health care delivery. Educational programs use many related terms, including bioinformatics, biomedical informatics, medical informatics, computational biology, clinical informatics, and others.
Master's degrees in bioinformatics are available at more than 70 institutions, and Ph.D.s at more than 30. The American Medical Informatics Association maintains a list of bioinformatics programs. It includes degree programs, certificate-granting programs, and programs targeted to health professionals. The list also includes postdoctoral fellowship opportunities and short-course and online learning options.
The U.S. National Library of Medicine supports 18 bioinformatics programs around the United States. The programs make special effort to recruit individuals from underrepresented racial and ethnic groups, individuals with disabilities, and those from economically, socially, culturally, or educationally disadvantaged backgrounds.
Nancy Volkers is a science writer in Vermont.