2012 Panel Discussion - Questions and Answers

Genomic Technologies and Focus

What non-medical uses do you foresee for genomic science and genomic information?

There are open-ended applications to basic science. For example, genomics is leading to a new synthesis of evolutionary, historical, anthropological, archaeological, and paleontological perspectives on human origins. When I was a young student, “ancient history” involved the building of the pyramids. Now, any telling of the human story spans millions of years and becomes quite rich over the past tens of thousands of years. Gauguin entitled a famous painting “Where Do We Come From? What Are We? Where Are We Going?” It is impossible to have an intelligible discussion of these fundamental questions without appealing to genomic data. Science is not entirely about achieving short-term practical benefits.

Genomics has also transformed the study of “model organisms” such as E. coli, yeast, flies, roundworms, mustard weed, and mice. Most of what we know about basic biology comes from laboratory studies of these organisms. Contemporary students cannot imagine how to do model-organism research without access to genome sequences and genomic tools.

Forensic applications of genomics must be considered a major success. Initially, the civil-liberties and criminal-defense communities were highly suspicious of this new forensic tool. However, DNA forensics has proven a potent weapon for identifying—and in some cases attenuating—abuses of the criminal-justice system. Not only have many innocent people been exonerated of horrific crimes, but prosecutors and judges have learned that much evidence on which criminal prosecutions have traditionally depended should be viewed with skepticism.

There has been a lot of interest in “synthetic biology,” the idea of going beyond gene-by-gene, or even pathway-by-pathway, genetic engineering of new organisms with potentially useful properties. I am a skeptic about both the hazards and benefits of this modern embodiment of our Frankensteinian impulses. Nonetheless, genomic data and genetic engineering are having major effects in agriculture and will undoubtedly play a role in developing sustainable energy sources.

Conservation biology is a growing application area. For example, in China, sophisticated genomic methods have led to improved estimates of the number of giant pandas in the wild (encouragingly, these estimates are larger than those based on traditional field observations). More importantly, genomics is the key to managing the highly fragmented populations of wild pandas. To maintain genetic diversity, it will be necessary to transport occasional animals from one preserve to another: genomic data and population-genetic theory will play essential roles in guiding this process.

These examples are just a small sampling of non-medical applications.

Genetically modified food remains a controversial and poorly understood issue. Intellectual property, allergy, and ethical concerns have all been raised. What do you see for the future of this field?

I addressed aspects of this complex issue above. This question correctly embeds the technology in a daunting web of non-technological issues. The future of the field must respect the reality that the technology is only part of a larger puzzle. We should change the way we train students to work in this area: the need is not for more narrowly trained technocrats but for scientists who are conversant with the whole range of issues posed by genetically modified foods.

How far are we from creating an organism – likely a microorganism – with a fully designed genome?

We must distinguish between the illusion and the actuality. Indeed, by permissive definitions, designer genomes are already a reality. I think much of what create-brand-new-life-form enthusiasts have to say is silly. We have no idea how to design a genome other than plagiarizing evolutionary experience.

Large sums of public money have gone into GWAS and now exome/genome sequencing in search of causes of common diseases. Is it, has it, been well spent?

GWASs (Genome Wide Association Studies) are a particular method of doing case-control studies aimed at identifying common variants in the human genome that affect specific traits, primarily disease susceptibilities. Posing the question of whether or not GWAS was a bust is a reliable way of starting a bar-fight at a genomics meeting. I think it was a bust, and a largely predictable one. Some of its enthusiasts consider it a great success. Claims of success take two flavors. One argument emphasizes the hundreds of genes, variation in which has been shown—probably fairly reliably, although there is controversy on this point, too—to influence disease susceptibilities; and, while admitting that the discovered effects are small (rarely accounting for more than a percent or two of the variance in disease susceptibility), this argument predicts that future research on GWAS-discovered genes will yield much medically useful knowledge. Another argument, coming to the fore during the past year, argues that GWAS effects were actually bigger than commonly recognized once putative fallacies in standard statistical methods are taken into account. My view is that geneticists have never gotten anywhere by analyzing small effects. Geneticists make progress by discovering particular situations in which genotype-phenotype effects are large and biochemically tractable, not by chasing after every tiny influence on phenotype.

Exome/whole-genome sequencing is in its infancy. However, this newly practical approach is already driving the discovery of rare variants of large effect. Given the complexity of human biology, we will need a lot more of genome sequencing—the future clearly lies with this approach.

The inability to predict phenotype from genotype seems to be a persisting weakness in genomics. In light of this fact, do you predict a shift away from broad, genome-wide research back to focused, hypothesis-driven research?

I comment above on the likelihood that there will be a movement away from efforts to predict phenotype from genotype. Genome-wide research will drive the discovery of new genetic phenomena; however, the choice of which of these phenomena to pursue, and the approaches used to pursue them, will fit more comfortably with traditional ideas of hypothesis-driven research.

The last era was Genetics – the next is Environment. Please comment.

I am a skeptic about our ability to understand—and even the medical relevance—of subtle environmental effects. The easy gains in environmental manipulation involve vaccination programs, clean water, and a safe supply of nutritious food. The next level of intervention requires changing human behavior. We already know where the big opportunities lie: smoking, diet, and exercise. There has been some progress with smoking, at least in the U.S., but mostly we have learned how hard it is to change people’s behavior. Subtle, and often idiosyncratic, environmental effects are exceedingly difficult to quantify. The goal is worthwhile, if a new era of environmental medicine is dawning, I have missed the signs.

What role does synthetic biology /genetic engineering have for preserving biodiversity in the present/future?

I briefly address this question above in connection with genomic applications to conservation biology. Genomics is a powerful tool for doing environmental inventories and, in some circumstances, for managing stressed populations. Primarily, genomic data will improve our ability to model the effects of various threats to biodiversity. How we use this information will be driven by politics and economics.

Given its complexity and heterogeneity, is cancer a useful focus of genomics?

Cancer is an excellent focus for genomics. The ability to compare tumor genomes with those of normal cells from the same individual allows identification of potential “driver” mutations. Sometimes the signal is lost in the noise of many functionally irrelevant genomic changes, but clear patterns do emerge for many cancers. Often these changes allow more precise classification of the tumors; less often, there is a direct path from these molecularly based classifications to effective, individualized treatments. That said, progress remains slow for several reasons. As the question indicates, by the time tumors present clinically they have often undergone immensely complex genetic changes and the population of malignant cells has often become highly heterogeneous. Most clinical successes of genomics have involved tumors whose genetic properties are at the simple end of this spectrum.

To take better advantage of the opportunities, we need a shift in the drug-development paradigm. We need a large panel of drugs that inhibit signaling pathways that are aberrantly regulated in cancer. In an ideal world, genomic monitoring of tumors would allow the use of combination therapies tailored to the individual patient. Presently, too few drugs are available, and it is extremely difficult to carry out clinical trials with combinations of the drugs we do have: often, different companies control access to the drugs that researchers would like to combine in experimental drug cocktails. Finally, most cancer patients are still treated with radiation and chemotherapeutic agents that act by damaging DNA. We have little understanding of how to tailor these broad-spectrum therapies based on genomic data. For that matter, we understand poorly why these agents work spectacularly in some instances (e.g., the use of cis-platin in testicular cancer) and not at all in many others. In general, pediatric and germ-cell-derived cancers are more treatable than the typical cancers of middle and old age, probably because they often present clinically before they have evolved high levels of genetic complexity and heterogeneity. In short, cancer genomics confronts enormous challenges, but there has been enough early success to justify major investment in this area. We should be patient when things are going our way.

What role will the crowd play in the future of genome science, with tools like Foldit, Galaxy Zoo, Big Data, Problems, etc.? Will crowd-sourcing solve problems, especially in enabling new understanding by crowd funding new projects, as the barrier to entry decreases and technological power increases?

I have addressed aspects of this question above. We absolutely need to engage the crowd. I think the main mechanism will have to involve integration of research with the ordinary course of health care for millions of individuals. However, for most of these people, engagement with the research system will be largely passive. Opportunities for active engagement will also abound, particularly in the case of rare diseases. Disease-interest groups are already self-organizing, sharing information about their genomic profiles, and driving increased attention from the research community. More power to these groups!

As genomics leads to understanding complex human disease (if it does), will there be an important role for systems biology?

Assessments of “systems biology” vary widely, as do definitions of this field. Without question, the future lies in integrating information, as quantitatively as possible, across biochemical pathways and whole networks of pathways. In my view, it is difficult to see solid successes with this approach so far. When I read a “systems biology” paper that does reach solid conclusions, I often have difficulty seeing what was particularly distinctive about the approach. On the other hand, in papers that describe work that clearly did take a “systems” approach, I often have trouble finding the solid conclusions. As biology continues to become more data-intensive and more reliant on modeling methods pioneered in other fields, this situation will undoubtedly change. However, I expect real change to occur slowly, precisely because biology has such high intrinsic complexity. Several years ago, I chaired a National Research Council committee that issued a report with the grandiose title “Mathematics and 21st Century Biology” (http://www.nap.edu/catalog.php?record_id=11315). This report takes a broad look at the potential for applying mathematical methods in biology at all scales of biological function, from the molecule to the ecosystem. It documents historical successes and points to promising current developments and future opportunities. However, the current developments are more in the “green shoots” category than a big surge of new growth.

Where do you see Genome Science 5 years from now?

See my answer to the last question in the set of questions actually discussed by the panel.

Regarding moving from data/description to actual understanding as mentioned in the introductions: What direction in genomic sciences do you think are most promising or will you be following?

This transition depends on a shift in focus from the big picture to some detail that merits careful investigation. We need to improve our ability to make such selections and also increase the efficiency of the methods we use to pursue particular phenomena in depth. I already see progress in both areas. Historically, human geneticists have often studied particular families for years or decades, eventually identified a mutation that accounts for the clustering of genetic disease in the family, and then struggled to develop “actual understanding” of the situation. Although this paradigm has served medical genetics well, it needs to change. Any model-organism geneticist knows that obtaining mutations that affect a particular phenotype is the easy part of functional analysis. The hard part is identifying the handful of instances that are highly informative about mechanism. Genomics needs to learn from this lesson. Genomic methods allow surveying large numbers of patients for potentially causative mutations. Now we need to develop better criteria for choosing the small number of these aberrant genotypes that offer particularly promising paths forward. We also need to develop a larger repertoire of efficient research tools with which to carry out the functional studies. In the latter area, there has been dramatic progress (e.g., the use of RNAi methods, improved gene-knockout, gene-tagging, and protein-tagging technology, GFP, opticogenetics, better methods of immortalizing cells). However, biology is complicated. We should never be complacent about the tools we have, no matter how powerful they might seem. Progress in biology is almost always driven by technical, rather than conceptual, advances.

As the nascent paradigm of “P4 Medicine” (predictive, preventive, personalized, and participatory) emerges and matures, how will the role of, and the need for, physicians change? Will they perhaps go the way of scribes in the wake of the printing press?

No. There will be more disease in the future, not less, and the disease processes will pose ever-increasing medical complexities. These complexities will far outstrip whatever progress Apple makes in teaching Siri medicine, thereby leaving a huge role for human physicians. Success in medicine, of which we have had a lot and will see much more, increases the burden on the health-care system rather than decreasing. The reason is that with better medicine people live longer and die of more complicated causes. Pneumonia was once known as “the old man’s friend” since it killed so many of the elderly before they slid into multi-factorial sources of disability. Dealing with this paradox poses enormous social challenges. The solution cannot be to allow people to die early of preventable causes. However, it also cannot be to do everything medically possible until the last minutes of life. In between these options, we will need plenty of physicians to provide expert, humane, affordable care for our aging populations.