Relying on Computers

Given the incredible advances in laboratory science in the last quarter of a century, it may seem that what the lab of the future will look like is anyone's guess. However, because those advances have included the ability to generate an astounding amount of data, a common theme for the lab of the future is continued, if not increased, reliance on computers. For Katherine Peterson, a staff scientist at the National Eye Institute (NEI) at the National Institutes of Health (NIH), Bethesda, MD, the lab of the future will need to do a better job of multi-disciplinary integration. “You can't putter away by yourself anymore in the present mode of data collection,” she observes. Integration of teams of people in day-to-day operations will be more to everyone's benefit. She believes that people in some disciplines, such as statistics and biology, are currently too far removed from each other. She's seen the shift from hypothesis driven research to projects that generate “piles of data. What's there, is the question,” she says. “Making the high-throughput information useful hasn't been done to its potential yet.” As the technology to generate data gets better and cheaper, she asks, “How will we cope?” Peterson would like to see more grant funding for large, interdisciplinary projects that would include scientists, mathematicians, statisticians, and computer scientists.

Community and Standards

Edward C. Uberbacher, Program Leader for Computational Biology at Oak Ridge National Laboratory, Oak Ridge, TN, agrees with Peterson. “From my perspective, the only way is to use a community approach.”

His group involves statisticians and mathematicians using a tera-scale, high-performance computing structure to design approaches to solve biologic problems. In collaboration with research groups collecting genomic, mass spectroscopic, expression, and other data, they investigate molecular systems to understand cellular pathways and regulatory circuits. They also use computing systems to analyze data from their own laboratories, other laboratories, and the literature, to build testable models. One “age-old” problem that complicates this process, which Uberbacher says predates his lifetime in computational biology (1990), and is ongoing, is the absence of universal standards for data collection for all of the types of information they use. Some data, he concedes (e.g. expression data) are hard to standardize. Although there are efforts, such as that of The Jackson Laboratory, Bar Harbor, ME, to standardize collection of mouse gene expression data, Uberbacher believes that due to the overwhelming amount of novel types of data being generated, the problem of standardization may not be solvable near term.

Image 1.

Image 1. (Click to enlarge)

Robots and Models

The Hospital for Special Surgery, New York, NY, is devoted exclusively to the study and treatment of musculoskeletal and autoimmune disorders. There, Timothy Wright, Senior Scientist and Director of Biomechanics and Biomaterials, and colleagues are using computer-aided surgery to develop and validate surgical navigation and medical robotics in orthopedic surgery. “Our work with computer-aided surgery is the lab of the future,” Wright says, “as more and more, we are taking advantage of technology to develop platforms for research.” His group's structural analysis of joints and joint replacements using cadavers is still relatively useful, he says, but has limitations, including difficulty obtaining sufficient material to study, and the wide variability of material due to age, disease state, and anatomy of the donors. Therefore, they are developing computer models for structural analyses. The analyses can be used to validate the computer models and allows them to perform bone strain and other calculations they could not do with anatomic specimens. “We are using different complementary technologies to give us confidence in our research,” Wright says.

In a shoulder function model, they can perform 30 computer model validation runs a day compared with one cadaver test per day. “These approaches complement each other,” says Wright. “The computational models have been coming into their own recently with an increase in computational power combined with decreased costs.” The group has also adapted robots like those used on automobile assembly lines to test knee joint function to develop navigation tools to aid anterior cruciate ligament (ACL) replacement. ACL tears are very common, and the ligament doesn't heal, so it must be replaced rather than repaired. The robotic model helps surgeons deter mine the best position to locate the replacement graft to allow the best joint motion.

Image 2.

Image 2. (Click to enlarge)

From Lab to Clinic

Wayne W. Grody, Professor, UCLA UCLA School of Medicine, Los Angeles, CA, directs molecular diagnostics and research laboratories and also sees patients. With the increase in automation using higher throughput technologies and higher density microarrays, he says “I worry on both the clinical and research side that there are more data than we know what to do with.” His clinical diagnostic laboratory does not use high density microarrays because they are performing targeted mutational analyses of common, recurring mutations of one gene with a limited set of mutations. Grody expects in the future, as more is learned, that targeted mutational analytic panels will get larger. For diagnostics for other diseases (e.g., cancer, atherosclerosis, and hypertension) high density microarrays will be required, but at this point in time, the identity of the genes, as well as their number, which will be large, are unknown. Grody has seen pharmacogenomics, which looks at genes for drug metabolism rather than for disease genes, evolve from a concept to reality. “What are those metabolic genes for?” he asks. “They must have some other purpose in the body. They didn't evolve to metabolize drugs that have been developed only recently.” He believes that if these genes are defective, they might have some associated disease process that has not been identified, so pharmacogenomics ought to be thought of as genetic testing. “I do worry about the ethical issues of genetic testing.” He thinks the $1000 genome sequence will be available fairly soon, but will “do more harm than good because we can't handle the data yet.” In spite of that, he would like to see more genomic analysis to determine which differences are polymorphisms and which are true mutations.

Image 3.

Image 3. (Click to enlarge)

Image 4.

Image 4. (Click to enlarge)

Well in Hand

Mohammad Al-Ubaydli, author of Handheld Computers for Doctors, and a physician and programmer, is interested in how the personal digital assistant (PDA) improves productivity. Although a computer, the PDA is limited by its storage capabilities and small screen. Therefore its strength lies not in data processing but as a data collection tool, whether bedside or in remote areas. Assets include a long battery life compared with that of a laptop or notebook computer and a global positioning system (GPS). Although PDAs are used increasingly in developed countries for clinical trial data collection, to access medical records, and to aid in bedside clinical decisions, one of the most successful public health uses is in developing countries, where PDAs are cost-effective and allow more data to be collected and disseminated more quickly than when paper is used. Al-Ubaydli notes that PDA technology allows organizations like the Centers for Disease Control and Prevention and the World Health Organization to track diseases and potential epidemics worldwide.

Time Management

Computer technology in the form of a web spider helps Peterson search for additions to the list of associated genes and diseases she keeps, something she wouldn't do otherwise. Al-Ubaydli had noticed at several scientific meetings that representatives of the Nature Publishing Group referred to their podcasts as their most popular feature. “I thought it was silly,” he says. “What's the use?” But he realized that it was possible to “keep up to date by having the journal read to you while doing mundane and necessary tasks” in the laboratory. He thinks that other journals are watching what Nature is doing and expects to see more scientific podcasts in the future.

Future Directions

Project Kaleidoscope (PKAL), an informal alliance supported in part by the National Science Foundation, the ExxonMobil Foundation, the Fund for the Improvement of Postsecondary Education, the U.S. Department of Education, and the W.M. Keck Foundation, has as its goal building strong learning environments for undergraduate students in mathematics, engineering, and various fields of science. PKAL suggests that this future generation of scientists, who will have grown up tech savvy in an electronic world, will find that the lab of the future will be, among other things: (i) digital, relying on ever-increasing computer power; (ii) small, as nanotechnology allows smaller instrumentation requiring smaller samples; (iii) virtual, where computers in concert with visualization devices turn both real and simulated data into interactive images, enabling users to overcome both scale and time to “see” (e.g., intracellular molecular processes); and (iv) openly collaborative and remote (e.g., by accessible databases, teleconference, and the Internet) and relying on linked computer systems, so anyone anywhere can interact in real time. Whether this proves to be the case, and what future scientists make of the lab of the future, remains to be seen.