Name: Edison Liu
Title: Executive Director, Genome Institute of Singapore
Professional background: 2001-present, executive director, Genome Institute of Singapore; 1996-2001, director, division of clinical sciences, National Cancer Institute; 1993-1995, associate professor, departments of medicine and epidemiology and biochemistry, University of North Carolina at Chapel Hill; 1987-1993, assistant professor in medicine and oncology, University of North Carolina at Chapel Hill
Education: 1983-1987, postdoc, department of microbiology, University of California at San Francisco; 1982-1985, hematology fellowship, University of California San Francisco, Moffitt Hospital; 1980-1982, oncology fellowship, Stanford University; 1979-1980, residency, Barnes Hospital, Washington University, St. Louis; 1978, MD, Stanford University; 1973, BS, chemistry, Stanford University
For life science tool vendors, a primary focus for new technologies is their uptake in the marketplace. But for Edison Liu, Jackson Laboratory's newly appointed president and CEO, genomic research tools like microarrays and sequencers are part of a more vast equation composed of understanding the components of human disease and making them clinically actionable.
Replacing Rick Woychik to begin his tenure at the Jackson Lab in January 2012 will require Liu to step down as executive director of the Genome Institute of Singapore, which he co-founded in 2001 and has since led.
During his time in Singapore, Liu initiated and managed the Singapore Cancer Syndicate, a funding agency designed to enhance clinical translational oncologic research; was executive director for the Singapore Tissue Network, the nation's tissue bank; and was an early member of the Bioethics Advisory Committee, which advised Singapore's cabinet on matters relating to research ethics.
Huck Hui Ng, associate director of the Genome Institute of Singapore, will take the reins when Liu leaves.
In the meantime, Liu is considering how to parlay those experiences into his new position in Bar Harbor, Maine-based Jackson Lab. Among his goals is to see the facility move beyond its expertise in mouse model systems and into human genetics by using genomic technology to improve the lab's ability to translate its research into clinically useful tools.
BioArray News spoke with Liu about his new job and keeping pace with what he calls the "fast-moving" genomic technology landscape. Below is an edited transcript of that interview.
Why did you decide to leave GIS for the Jackson Laboratory?
I've been in Singapore for 11 years now and I think we have accomplished what all of us had set out to do, on a professional, personal, and national-development level. We built a program that we can be proud of, recruited the right people, and gave them the opportunities to make a difference in the field. Internally, we have built infrastructure [that is] quite respectable with [the help of] a pipeline of young talent that is Singaporean who can actually now take over.
Besides that, it's always been my belief that any founder of an institute shouldn't stay beyond 15 years. Science moves very quickly now, and it is an issue of allowing new leadership to take programs to a different level. An institution gets stale if it has the same old guy running it. My planned obsolescence was already in the works, but it was accelerated a bit because of the opportunity with the Jackson Laboratory.
I have a fantastic young scientist, Huck Hui Ng, who is going to carry on the initiatives and sustain the culture we have at GIS. He's a Singaporean, and we are handing it off to the next generation of scientists who happen to be Singaporean, which is exactly what we wanted to do.
What would you like to accomplish at the Jackson Laboratory?
In talking to the search committee and to the board, and then to the scientists, it became clear that Jax has been extraordinarily successful in developing the mouse as a model system, and mouse genetics. They wanted to move more toward human genetics, to take their knowledge and seek some kind of translational interface with medicine and to take advantage of the genomic explosion, which is what they are beginning to do, but really wanted to accelerate in a different way. I sort of fit the bill for that new direction.
In addition to that, Jackson understands that for their business component they want to consider expanding out of North America in a globalized manner. Given my background, I have some expertise in that as well.
I would like to explore the interface between two complex organisms, human and mouse. To make things translational, from my perspective, one has to look at human health and human disease. Exciting for me is how one would use the mouse as a test bed for systems complexity at the organismal level and to draw parallels to the human condition. Hopefully, we will be able to deconvolute some of the complex systems that are seen in many human diseases.
Do you see any opportunities for increased cooperation between the Jackson Lab and Asian research institutes and organizations, in particular?
Absolutely. One of the things I am already talking to GIS and Jackson Lab is interactions for mutual benefit. GIS has a very robust systems-genomics infrastructure, but we have very little in terms of mouse. I can imagine my colleagues in India, and to a certain degree in China, where the interface may be extraordinarily helpful.
[ pagebreak ]
The Jackson Laboratory has focused to date on a number of research areas such as oncology and neuroscience. Would you like to expand the scope of its focus?
Oncology and neuroscience are two areas of excellence at the Jackson Lab, but the difference is that I would like to expand the human-genetics interface for both disorders in a manner that will bring human disease into those model systems that Jax has great experience with. Should we go into other areas? Of course. But that is dependent on the quality of people that we can recruit and in terms of key collaborations that we can strike up.
There was also a recently abandoned bid to expand the Jackson Laboratory into Florida. Will you continue the effort to open a research institute beyond Bar Harbor and Sacramento, Calif., where the lab has a sister facility?
We are, and we are doing it for strategic reasons. I think the experience with Florida is that we should pay some attention to complementarity. We are doing so on the basis that science needs connectivity. Just like GIS, our strength comes in the form of deep collaborations. In the case of Jax, some of that appears to be best done if there are known centers of expertise where we can have a physical presence. That was explored in Sacramento and turned out to be a great success. The focus there is really more service toward pharmaceuticals and biotech. And that has built an expertise around translational sciences, which is very interesting. So what we plan next will be dependent on the strategic interest of both parties and whether or not there's complementarity and resources and the will to have it done.
[Jackson Lab] will never abandon Maine, and I hope to expand our presence there, but to get involved in clinical genetics we need strong academic and clinical partners, and Bar Harbor is very distant from some clinical academic institutions. We would be interested in looking elsewhere for those possibilities.
Other goals of the Jackson Lab are personalized medicine and translating research into medical practice. How can the facility improve on these?
I think the concept of personalized medicine is a fundamental [goal] of modern medicine — to better identify groups based on some characteristic of the patient that we can tailor the therapy to increase efficacy and reduce toxicity. That's called the therapeutic index.
The difference is two things: One is the precision and comprehensiveness of the technologies we have and assessing the uniqueness of each individual, so that we can identify people who benefit from one approach over the other. The second is that we are dealing with complex combinatorials. In the past, we would take one target and attack it. We now know that if you do that, there are unintended outcomes. And despite the fact that a number of these compounds hit the targets, they don't always have efficacy.
The question is, 'How do you identify individuals where they will be most efficacious?' We are really moving into this concept of systems medicine, where we can identify all the complement parts in the system and address how these parts interact so we can at least reduce the outcome space. In which case, we can talk about combinatorials in therapies, and actually altering doses in combinations to get the optimum downstream effects. That systems reconstruction of disease and narrowing of outcomes is something that we can accomplish with the personalized medicine we are talking about.
Now, fundamental to this is the ability to identify all the complement parts so that we can start that reconstruction process, and I do believe that genomics, the reduction in sequencing costs, and the ability to analyze the output of the various sequencing components is going to allow us to do that kind of reconstruction.
Are the channels in place now to make your findings actionable?
Actionability is going to be a challenge. The question is, 'How do we parlay a mouse system into this?' First of all, it is quite clear that the genomic-technology interface, including arrays, sequencing, and the analytics, is vital for that interface. Here, partnering with the appropriate clinical colleagues is going to be critical.
Can you provide an update on Pan Asia SNP? I have seen a few population-specific papers in recent years. At what stage is that initiative, and where is it going?
That project never had as its primary target a study group that would expand and continue forever. It was really an opportunity for colleagues to get together and do something scientifically and organizationally unusual: to launch something that was conceived, funded by, and executed by Asian genetic institutions.
What we are talking about now is the second phase. It has to be different. The technologies, the questions are different [and] will inevitably engage deep sequencing. At our last HUGO meeting, we had several conversations about identifying the genetic borders of Asian populations and asking with deep sequencing the difference and similarities between these individuals.
We haven't solidified the plan simply because right now most of us are interested in doing something impactful together [rather] than just doing something. It wasn't clear to us that sequencing 30 or 40 distinct Asian populations was going to be helpful given the 1000 Genomes Project experience, and how many centers are doing genome sequencing. But at the Sydney Human Genome Meeting [in March 2012] we are going to have two special sessions on genomics of indigenous populations, and we hope to have a dialogue with leaders of these population groups to see if the science of genomics can allow them to look into their genetic ancestry and for the scientific community to address ancient populations. These are things being discussed, but we don't have anything solidified at this point.
Some microarray vendors are developing population-specific or population-themed arrays for association studies. What do you think of these kinds of tools?
These population-based arrays may be a moot point because the technologies are getting so remarkable. We are already talking about a 5-million-SNP chip. Once you get to that level, you are kind of covering most populations, and also progressively rarer alleles. I think that when you have limiting technologies, these population-specific arrays are important, but as our technologies get better and better — for example, if we get to the $1,000 genome sequence, why not sequence 1,000 individuals? In that situation, you don't need population-specific arrays because you just sequence everybody. So it is really a technology basis. Technology is fast advancing and the relevance of that question is a bit of a moving target.
How do institutions like GIS or the Jackson Lab decide which next-generation sequencing platform to buy or which array system to bring in house? How does one integrate the newest technology into what one is doing, but not blow the budget on the newest toy and waste time trying to figure out how it works?
Well, you literally hit the nail on the head. This is kind of remarkable, because it wasn't that long ago that we didn't even have that option. It's probably not much different from what happened with computers [in which] every 18 months you had a new computer. The question for the IT managers was, 'Should we buy something now, or wait a few years for the next technology to come on board?' Before we think about it, we have to ask the question, 'What are you using sequencing for?'
Obviously, if a new platform isn't any better than the old one, there is no reason to go for it. And buyers are becoming more discerning. They are going to vendors and saying, 'Prove to me that what you do makes sense for me.' As you know, many vendors now do test runs for the scientists to show them that the results are good, reasonable, and better than what they had.
And this amount of competition is driving the cost of sequencing down at a very fast clip, which is exactly what science needs and what scientists want. In this competitive situation, the best technology will win.
There are two forms of sequencing that one will have to deal with. One is what I call horizontal sequencing. I have one thousand genomes I have to sequence in the same way. Most institutions can't do that. So you rely on BGI and Complete Genomics and a few other vendors who are willing to have that kind of capacity. But there is also the vertical capacity, where I need to answer my question in a most cost-effective manner. In that setting, you need vertical capabilities, [and need to be able to] generate new libraries and analytical tools along the way. Both formats will be needed and require a different framework. For horizontal capacity, people are outsourcing, for the vertical capacity, institutions are keeping some of that expertise in house.
We have talked a lot about sequencing. Is there room for arrays in this mix?
Yes, and I believe there is no question about it. But for arrays to be competitive they have to be cheap and the analysis has to come easily. There is no question, whether it’s a metabolism chip or a cardiovascular chip, there is going to be a need for fast screens of a limited number of loci, not the whole genome. In order for that to be competitive, it has to be inexpensive, and driving the cost down is going to be important.