Skip to main content
Premium Trial:

Request an Annual Quote



With all the talk about how systems biology is going to change the life sciences industry, ever wonder what it could look like in action?

For a clue, take a tour of the 11- month-old, state-of-the-art Novartis Institute for Biomedical Research in Cambridge, Mass. Pass through functional genomics and chemistry labs, scribble your thoughts on dry-erase glass partitions, hang out in an auditorium listening to a lecture about the newest advances in biomedicine, or have lunch in bamboo-floored break rooms with biologists, chemists, computer scientists, perhaps even engineers. Then wander up to the top floor in vitro biology labs and gaze out on the rooftops of some of the world’s leading academic institutions of biomedical research: the Whitehead and McGovern Institutes, and the future site of the Broad Institute. This, the theory goes, is the atmosphere that will nurture the sort of scientific discovery that will revolutionize biomedicine.

It’s not news that systems biology is being heralded as the future of biology. From Lee Hood’s founding of the Institute for Systems Biology four years ago, to the sprouting over the past two years of a dozen major academic systems biology programs, to NIH’s recent promise that it will give top funding priority to interdisciplinary biomedical studies, it’s clear that systems biology has taken hold. By linking up the modern tools of biology — genomics, proteomics, computer science, engineering, and so on — the current generation of biologists will begin to understand the networks of living systems.

The concept has also sparked the interest of the investment community, with venture capitalists backing several startups that aim to provide systems biology services to the pharmaceutical industry.

But is big pharma up to the task of using systems biology to transform drug discovery? In spite of efforts like the one at Novartis to create breeding grounds for systems biology, some say that traditional pharmaceutical companies lack the organizational flexibility to make it work. Industry observers say big pharma may be able to integrate data from various biological experiments, or extract patterns from genomic data, but they doubt that systems biology will flourish there. That’s because the ultimate goal of systems biology — constructing models of complex biological systems that would enable researchers to predict the outcome of particular experiments — may be beyond the ken of the highly structured behemoths.

Nat Goodman, a bioinformaticist at Hood’s Institute for Systems Biology, offers a blunt critique: “Honestly, I don’t think big pharma has a prayer of pulling this off. They’re going to waste a lot of money on this, but their organizational structures do not allow them to create the interdisciplinary teams they need. It’s a good way for them to spend their remaining cash.”

But big pharma sure is trying. While some drugmakers haven’t done much more than slap the systems biology label onto existing genomics or proteomics programs, several are embracing the new approach to discovery. Eli Lilly, with a group dedicated to systems biology in Singapore, has arguably the most well-known effort, and others, including AstraZeneca and GlaxoSmithKline, are making smaller but equally focused attempts to keep themselves on the cutting edge.

“In some pharma companies they recognize that doing this integrated thing may require them to start a completely separate activity that has the title of systems biology,” says Robert McBurney, CSO of Beyond Genomics, a startup based in Cambridge, Mass., that describes its multi-pronged approach to drug discovery as systems biology. Other pharmas, he says, “are giving the systems biology title to one of the existing groups and essentially giving them much more power to pull things in.”

A phrase to fit any notion

Many would agree that boiled down to its ultimate goal, systems biology means combining biology — in all its levels of complexity — with mathematics and engineering to create simulations useful for predicting how biological systems will respond to certain stimuli.

“What we’d like to be able to do initially is create graphical models that integrate many different types of data, and ultimately mathematical models that describe the behavior of the system,” says ISB’s Hood. “Once you can do that, you can explain the emergent properties of the system, you can predict its behavior given even new perturbations, and most important of all for medicine in the future, you can redesign the system to create new emergent properties or to change in some fundamental way its properties. In this case you can do that with drugs.”

Yet it should come as no surprise that many researchers interpret Hood’s definition to suit their own purposes, particularly within large pharma companies. At Indianapolis-based Eli Lilly, the company’s initial concepts for applying genomics to drug discovery have since expanded into a dedicated systems biology program, in which bioinformaticists consolidate and analyze biological data taken from a wide variety of platforms, says Harry Harlow, the scientific leader in systems biology informatics who oversees all of the bioinformatics scientists at Lilly.

Together with the Lilly Systems Biology group in Singapore, a research arm established with partial support from the Singapore Economic Development Board, bioinformaticists in Indianapolis and from other Lilly research sites around the world collect data from gene expression, proteomics, transgenic knockout, genotyping, NMR and even clinical experiments, in an attempt to integrate and extract patterns from the data. “We view systems biology informatics as being the great integrator,” Harlow says. “We bring biologists, computer science capability, and bioinformatics all together to help work on specific projects.”

Among other projects, the systems biology group is helping scientists interpret the results from gene expression experiments. In general, Harlow says, such projects work like this: Lilly managers first assemble a team of researchers taken from the group studying the disease under investigation, the gene expression group, and the systems biology group. After designing the experiment and collecting the data, bioinformaticists in the systems biology group perform the statistical analysis, integrate and annotate the significant results, and import the data into Spotfire’s DecisionSite software to produce an interactive report.

“The key is to have an interactive system that can be used as a communications tool between our bioinformatics scientists and the experimental biologist to solve real problems,” he says. Typically, Harlow adds, there are a number of such teams working simultaneously on different projects.

Harlow, who joined Lilly in early 2003 after serving as director of bioinformatics at Monsanto, says his systems biology teams have also developed their own methods for teasing out patterns from genomic and other biological data. In addition to cluster correlation analysis of gene expression data, Harlow says his team applies other algorithms for pattern recognition, including principal component analysis and correlation image mapping. The team of bioinformaticists in Singapore, he adds, are also engaged in devising new techniques based on non-parametric data analysis, and are incorporating technology from chemical engineering, signal processing, and applied physics.

In the future his group hopes to develop a systems biology database that would allow researchers working on a new therapeutic agent to ask questions about what is currently known about a biological model, pathway, or individual target. To do this, Harlow’s team plans to integrate classical biology — what he describes as phenotypic data — with clinical measurements and data from high-content genomics experiments, such as metabolomics, transcriptional profiling, and proteomics. “Having data ‘worthy of analysis,’ together with new advanced analysis methods and a systems biology database, is expected to help our group improve [Lilly’s] efficiency in developing new pharmaceutical products,” Harlow says.

At AstraZeneca…

Otto Ritter, associate director of bioinformatics at AstraZeneca’s Waltham, Mass., research site, is following a similarly diverse approach to bringing systems biology into the drug discovery process. In addition to embedding bioinformaticists in various disease-oriented teams, Ritter’s group is engaged in developing shared data resources and tools that the various therapeutic teams can access remotely. And his definition of systems biology jibes closely with Harlow’s — that systems biology is the glue holding the various pieces of the discovery process together.

“I would define systems biology as the mapping and navigating of the multitude of spaces [inherent to drug discovery],” Ritter says. “Communication between mathematicians and biologists is difficult, but the more shared experience, the better they communicate. With systems biology, you can represent different views in one model.”

One of the ways Ritter’s group is trying to stitch together the experiments and analysis required for drug discovery is by creating a database of molecular interactions and associations accessible to AstraZeneca researchers across the globe. The task, he says, is to combine data from yeast two-hybrid and mass spectrometry-based immunoprecipitation experiments performed at AstraZeneca with publicly available interaction data stored in such repositories as BIND. Ritter says he can avoid tackling complicated questions of database architecture by keeping the format simple: “The database [design] is not a big issue — you just treat it as a network,” he says. “Domain types are represented at the level of data contents.”

Partly because AstraZeneca’s cancer discovery efforts are headquartered in Waltham, Ritter’s group has also begun working with academic collaborators in the Boston area to create models of cell-signaling networks. In one project, Ritter is working with Doug Lauffenburger, a chemical and biological engineer at MIT, to understand the mechanism of action of Iressa, AstraZeneca’s recently approved drug for treating non-small cell lung cancer.

In an effort to determine why Iressa is more effective in some patients than others, Ritter and Lauffenburger, together with a postdoc from Lauffenburger’s lab working at AstraZeneca, are performing a series of cellular assays to collect data on how certain types of cells respond to stimuli. In parallel, the team is using the biochemical data to refine computer simulations, involving sets of ordinary differential equations, that the researchers hope will lead them to new hypotheses, and ultimately to an understanding of how to predict which patients will respond to the drug.

Ritter says systems biology at AstraZeneca also incorporates his efforts to design a comprehensive text-mining platform. Writing software to pick out specific molecules by name or type from paper abstracts would help link the pharmacology and chemistry of drug discovery with the underlying molecular biology, and potentially provide clues to the safety and efficacy of new drugs under investigation, he says.

Glaxo and Novartis outsource…

Other pharma companies have taken their lead in systems biology from collaborations with biotechnology companies. Pfizer is working with Entelos, a company that builds physiological models associated with specific diseases, and Novartis is working with Beyond Genomics, a company that classifies its current attempts to combine data from gene expression, proteomics, and metabolite analysis in the search for biomarkers as systems biology. Although neither Novartis’ Cohen or McBurney at Beyond Genomics would disclose the exact nature of their work together, McBurney describes his company’s general approach as that of using statistical and pattern-recognition algorithms to sift through data generated by multiple platforms and identify patterns associated with disease. The company is not currently engaged in modeling cellular systems, but claims that its ability to integrate data from mRNA, protein, and metabolite measurements gives researchers deeper insight into cellular function.

GlaxoSmithKline is also experimenting with systems biology through a collaboration with Beyond Genomics, says Michael Luther, Glaxo’s vice president for high-throughput biology discovery research. Luther’s definition of systems biology includes finding new ways to use in vitro or animal-model analyses to predict the effects of a treatment in humans, and his company’s work with Beyond Genomics is one part of this effort, he says.

In their work together to date, Beyond Genomics and Glaxo scientists are collecting gene, protein, and metabolite expression data from a clinical trial of an approved, marketed compound (Luther declined to divulge the name of the compound), and comparing the data with an identical study using an animal model. The goal, Luther says, is to identify the set of markers in the animal study that overlap with markers from the clinical trial for a given therapeutic outcome. Validating the markers identified in the animal model should enable researchers to more accurately predict how a drug might fare in humans, thereby “increasing the confidence in the compounds we take to the clinic,” he says.

And Aventis Collaborates with Academia…

At Aventis, Bob Dinerstein, a bioinformaticist responsible for applying systems biology to drug discovery, sees the approach as the next great consolidation of biological knowledge, and thus the key to new drugs. “From a pharmaceutical standpoint, it means we can assemble larger and larger data sets, and we need to learn how to integrate all that information with physiology to make better predictions,” he says. “There’s no question it’s where biology is headed, but the question is, ‘How do we get there from here?’”

Efforts at Aventis to predict the toxicity and bioavailability of lead compounds in humans “can be considered systems biology,” Dinerstein says, and more recently his group and others at Aventis have begun working with academic researchers at Rutgers University and with Entelos. As part of a collaboration with Rutgers’ BioMaps Institute, which bills itself as a quantitative biology program encompassing such research efforts as computational biology, mathematics, and biophysics, Dinerstein has recruited a BioMaps postdoc to work with his group in Bridgewater, NJ.

But like many of his colleagues at other companies, Dinerstein, a former researcher in the department of pharmacological and physiological sciences at the University of Chicago, suggests that it may be some time before systems biology begins to pay off for big pharma. Aventis has to incorporate systems biology “in a more validated manner” because the company must be careful to make sure systems biology adds value, he says. There’s currently no easy way for pharma to implement the approach to systems biology espoused by pioneers such as Lee Hood, he adds. “We’re looking to see how that proceeds, but we’re waiting to see when the hand-waving goes away.”

Why they may never get there

If and when the hand-waving does go away, it’s not clear that researchers within pharma can build the kinds of programs capable of accomplishing the ultimate goal of systems biology: to tie molecular biology with physiology through quantitative, predictive models. The barriers, industry watchers say, lie primarily in the research culture within big pharma that encourages silos in the R&D operation.

Traditionally, big pharma has tended to group researchers based on their location along the classical drug discovery pipeline: A target discovery group identifies a list of candidate gene or protein targets, then passes them along to a chemistry group, which in turn passes along a set of lead compounds to a group responsible for early-stage clinical trials. “The real challenge for the industry is whether they can do the integration necessary for systems biology,” says ISB’s Hood.

To be fair, managers within pharma say they’ve made progress in encouraging researchers with disparate areas of expertise to work together. At Lilly, Harlow says he tries to motivate his teams of biologists and bioinformaticists by taking an evangelistic approach to touting the goals of a project, rather than dictating the necessary tasks. At Aventis, Dinerstein says that it’s actually easier for him to create interdisciplinary teams than it would be in academia, where researchers are often motivated by self-preservation. Interdisciplinary teams form the “basic innovative units” of the company, he adds.

Hood and others are not so convinced. Fostering the scientific creativity necessary to do systems biology requires that researchers have “the freedom to operate without being second-guessed by those higher up on the ladder,” says Hood. He argues that to be effective, interdisciplinary teams need to work together eye-to-eye, not scattered across the globe at a pharma’s many research sites.

As a member of the advisory board to Lilly’s effort in systems biology, Hood says he’s critical of the geographic separation between biologists and computational scientists. “My deep warning is that you can’t separate the computational people from the biologists,” he says. “The biology really has to drive the process, and you can’t do that 7,000 miles apart.” Lilly’s Harlow argues, though, that working with researchers in Singapore can be advantageous because of the time difference. “It’s great in the morning when you come in and a lot of the work has already been done,” he says.

Peter Sorger, a molecular biologist and co-chair of the Computational and Systems Biology Initiative at MIT, believes that systems biology “if done correctly” will begin to have an impact on drug discovery within big pharma in three to five years, but that change may be incremental and will require realigning training and industry sociology “to break down the traditional barriers between the silos of knowledge.” In other words, Sorger says, “Think about the problem first and then adapt the social setting to it. Don’t say the sociology is a given; then you’ve already set yourself up for failure.”

If not in big pharma, then how?

It may not be necessary, however, for big pharma to do all the heavy lifting for systems biology to eventually make drug discovery easier. Given the number of new academic institutes sprouting up to foster interactions between biologists, computer scientists, and mathematicians, as well as pledges of NIH and NSF funding, there may be ample opportunity for public-sector efforts to work out some of the thornier issues in modeling biological systems first. A convincing proof-of-principle study showing the practical application of biological simulations to drug discovery would go a long way, many pharma researchers say, toward opening their doors to systems biology.

Industry-academic partnerships may represent the ideal model for translating the concept of systems biology into practical applications, say Sorger and Novartis’ Dalia Cohen. “It’s going to be a combined effort of the scientific community to develop the tools, to develop the hypotheses, so questions can be asked,” Cohen says.

“My guess is that it’ll be slow and difficult to make the necessary transition until we have enough cases where people could say, ‘You know what, that actually really did work,’” adds Sorger. The most convincing success story, Sorger adds, would involve applying the concepts of systems biology to understanding drugs’ mechanisms of action, and having physicians directly apply that information to the benefit of their patients. “If systems biology can have an impact there, then I think we will have established that in the short term it’s going to be useful to industry,” Sorger says.

In the meantime, many researchers are undeterred in their efforts to build the foundations for systems biology within big pharma. In fact, AstraZeneca’s Ritter believes one day he and his colleagues will be able to build a mathematical model describing the entire drug discovery process, a tool useful for gauging the probability that a given project will succeed in the clinic or on the market. “The process is distributed, and therefore inefficient,” he says. “One department is in charge of steering, and another in charge of the pedal. You can communicate by phone, and you would go from here to there, but it would take a very long time. If you have a model, it’s actually like driving the car yourself.”

Sounds great, and few would disagree that new ideas are drastically needed to aid the current approach to drug discovery. But will it be pharma that champions systems biology? Novartis certainly hopes its new research institute in Cambridge will prove big pharma can. How well it’ll succeed remains to be seen.



The Scan

Expanded Genetic Testing Uncovers Hereditary Cancer Risk in Significant Subset of Cancer Patients

In Genome Medicine, researchers found pathogenic or likely pathogenic hereditary cancer risk variants in close to 17 percent of the 17,523 patients profiled with expanded germline genetic testing.

Mitochondrial Replacement Therapy Embryos Appear Largely Normal in Single-Cell 'Omics Analyses

Embryos produced with spindle transfer-based mitochondrial replacement had delayed demethylation, but typical aneuploidy and transcriptome features in a PLOS Biology study.

Cancer Patients Report Quality of Life Benefits for Immune Checkpoint Inhibitors

Immune checkpoint inhibitor immunotherapy was linked in JAMA Network Open to enhanced quality of life compared to other treatment types in cancer patients.

Researchers Compare WGS, Exome Sequencing-Based Mendelian Disease Diagnosis

Investigators find a diagnostic edge for whole-genome sequencing, while highlighting the cost advantages and improving diagnostic rate of exome sequencing in EJHG.