San Diego’s role in computer analytics dates back to at least 1969, with the founding of Science Applications Inc., now SAIC, which used computer models to predict the effects of nuclear blasts for the Government. Today, a life sciences industry that didn’t exist in San Diego 40 years ago is embracing increasingly powerful computational tools to make sense of the massive amounts of data resulting from genetic discoveries and advances in molecular profiling.
Among the leaders in this emerging field is Paul A. Rejto, director of computational biology in oncology research at Pfizer in San Diego. A physical and theoretical chemist, Rejto started in 1994 at Agouron Pharmaceuticals, a San Diego biotech that became part of Pfizer in 2000. We caught up with Rejto by e-mail last week to find out more about the technology, its role in pharmaceutical research, and its importance to San Diego.
Xconomy: What is computational biology?
Paul Rejto: As with many interdisciplinary fields, there is no single well-established definition. Computational biology broadly refers to the application of computational and informatics approaches to address questions in biology. A number of other terms describe activities at this interface, with slightly different flavors. Bioinformatics is typically more focused on algorithms for sequence manipulation and analysis, and biomedical informatics is more focused on the acquisition and analysis of patient data and outcome.
X: What are some applications of computational biology?
PR: As you might imagine there are many applications. Here in the Pfizer Oncology Research Unit, we are applying computational approaches to support two major objectives: 1) identification and credentialing (or validation) of oncology targets, and 2) linking targets to patients using predictive markers of response. Our chief scientific officer, Neil Gibson, refers to these efforts as the bookends of the research portfolio – in other words, we support the selection of projects that are initiated within research and help to ensure a successful path for those projects moving forward out of research into the clinic.
To credential new targets, we use genomics and transcriptomics to assess targets and pathways in well-defined populations with unmet medical need, and look for evidence of functional activation. To accomplish this, we work closely with our commercial colleagues to ensure that we are addressing the most relevant disease opportunities, with physicians to obtain patient tumor biopsies, and we partner with groups that determine the molecular profile of the samples. To enable the generation of preclinical hypotheses for predictive biomarkers, we work side-by-side with wet [lab] biologists who perform in vitro and in vivo pharmacology on our drug candidates. Our group provides the computational framework to extract predictive hypotheses from their experiments.
X: What are the most promising future applications? And how far into the future will we see them?
PR: Computational approaches already permeate biology and there is a great deal of exciting work going on that has impact now. Speaking from the perspective of our work here at Pfizer, developing preclinical predictive biomarkers, translating these into the clinic, and then leveraging these markers to focus on patient populations that are enriched for likely responders permeates our research strategy, and has growing influence on the way in which we develop programs in the clinic. Combining the power of molecular profiling techniques including next-generation sequencing with well-characterized experimental preclinical models can be the difference between a successful registration [of a new FDA approved drug] and a failed clinical trial.
Moving forward, we are exploring opportunities to move beyond correlative analysis and increase our confidence in targets by establishing their causal relationship to disease. I think that it is critical to use computation as an enabling component of pharmaceutical research, but it is deeply misguided to consider the development of an abstract model as the objective of pharmaceutical research rather than the generation of novel chemical or biologic entities.
X: What needs to happen for this technology to reach its potential?
PR: Computational biology has many practical applications – our interest is focused on improving the abysmal and unsustainable low rate of new drug registrations. Simply put, we in research are not sufficiently productive to sustain the current pharmaceutical enterprise. The past ten years have been characterized by a series of business-driven efforts to incrementally improve productivity and profitability, such as outsourcing and the many mergers and acquisitions, but these have not fundamentally addressed the grand challenge we face.
Many of us are optimistic that the insight afforded by molecular profiling approaches will enable pursuit of targeted therapy more rapidly, effectively and efficiently. In addition to historical examples such as the Estrogen Receptor, HER2, and EGFR, there are many exciting trials of targeted agents in targeted populations currently underway at Pfizer and elsewhere.
I believe that the biggest outstanding challenges are not primarily technical, but rather cultural (e.g. while we are making progress, we need to dramatically increase communication and collaboration between preclinical and clinical scientists) and logistic (e.g. we need to accelerate the routine collection and broad access to consented tumor samples). Given that programs entering clinical trials have success rates under 10 percent, we may want to think more systematically about how to best leverage the information generated from the many trails that do not lead to successful registrations.
X: What is the role of the San Diego innovation community in advancing this technology? Besides fostering drug research and development, can the technology provide jobs and economic growth?
PR: San Diego is a hub of biotechnology, pharmaceutical research, diagnostics and information technology. Coupled with strong universities and research institutes, San Diego is well-positioned to play a significant role. Without question, the environment is far more challenging than when I started working here in the early 1990s. Successful companies today, be they small start-ups or major pharmaceutical giants, need to be smart, focused and lean. Those that are not are quickly punished.
Several of the recent hires performing lab work in the Oncology Research Unit have significant computational expertise, particularly in the area of statistics and scripting. Likewise, we look to hire computational biologists with experience and understanding of cancer biology. My recommendation to young scientists is that they combine deep computational expertise with real biological insight.