YouTube Facebook LinkedIn Google+ Twitter Xinginstagram rss  

Edison Liu talks about Singapore, HUGO, and predictive biology.

Feb. 1, 2008 | Edison Liu is Executive Director of the Genome Institute of Singapore, in many ways the flagship institute of the country. Liu manages the scientific direction, recruitment, and the budget for the entire institute. The institute employs some 260 people full time. Fifty percent are Singaporeans and 50% are foreign. This year, Liu adds president of HUGO to his many responsibilities.

Bio•IT World’s Allison Proffitt visited Liu in Singapore and asked him about the big picture of research in his institute, his country, and his new organization, HUGO.

On Singapore and the Genome Institute of Singapore

BITW: What is your focus moving forward  for GIS?

Liu: Our focus has been and will continue to be genome to systems, using transcriptional regulation and the transcriptome as the fundamental organizing principles. From there we seek a deep integration of technologies with biology that allows us to address very important biological questions that span from fundamental laboratory discoveries all the way to population studies. So that has been our leitmotif from the beginning and it’s continuing. The beauty, though, is that we have concrete examples of how that has worked extraordinarily well and that this integration has given us greater insights in biology and genomics than we normally would if people were working alone.

How is GIS different from other institutes?

First, we’re selective for people who want to do integrative science. Second, people are recruited for very disparate skills, whether you are technology or informatics or biology. You hope to think in biological endpoints, but you come from very different orientations. The challenge is that you don’t always have the same assumptions when you get into relationships or you get into a project. There we’ve set systems whereby individuals have degrees of freedom as individuals. But their advance is dependent on joining others. We have a process whereby if you want to do something new with new resources, you need to present that idea to the collective, your community. Everybody is welcome to come and listen. They comment and ultimately decisions are made in terms of whether a project is a go or no-go. And if it’s a go, what are the milestones that we want to monitor over time? We seek people and projects that drive convergence. That’s where the power is, when we kind of converge on the same question and arrive at the same outcome but coming from different directions. It’s really extremely powerful…

The fundamental attributes we reward are dogged adherence to excellence and, number two, a deep sense of collegiality and communal support. Now, those are actually opposing attributes. Okay, you have camaraderie and conflict. It’s something that, again it’s not easy to get at the balance.

On the Human Genome Organization

What is HUGO’s relevance and purpose?

That’s a very good question. HUGO was constructed and established to help governments know how to do the genome sequence. It’s been done. The challenge right now is really twofold. One is genomic medicine. All the things we’re doing right now are sufficiently mature that they’ll have an impact on how we diagnose, how we treat, how we develop drugs.

The systems biology concept at Lilly was completely a genome-to-systems approach. It was based on transcriptional responses. They’re using it as a very intriguing way to organize their drug development framework. 

To make personalized genomics real, we have to account for the interactive effect of groups of genes. That’s only one aspect. The genetic association studies show the multitude of genes or genetic loci that have small but significant impact on common diseases. It is the composite effect of many genes that will make up the population diversity contributing to a single disease. 

The speed in which this approach and genomic technologies are moving into medicine poses a tremendous challenge for the genomicist, the ethicist, the drug companies, and the patient. And it has to do with scale. When you’re dealing with small scale, there’s time, there’s all sorts of stuff that you can stop and go, but now it’s this rush that is 5, 6, 7 orders of magnitude faster and more comprehensive, and you’re flooded with information. How do you then dissect that all out? How we manage that, how we regulate that which always requires a computer intermediate is another challenge. How do we deal with a situation where I can know everything possible about you genetically? How do we safeguard it? What kind of legislative process?

Those are really important issues that an organization like HUGO can really contribute to… HUGO is the only [genetics organization] that has no geographic or national limitations.

And what’s the second part?

What we’re seeing since 1980 has been pretty remarkable — but in the last ten years has been dramatic — is a complete shift in world order. Not in terms of who’s on top and who isn’t, but in parity. The emerging economies of the world, from Korea to India, China to the southern shores of Australia, South Africa to parts of the Middle East, and Eastern Europe to all of Latin America, there’s really a dramatic change in both scientific capabilities, and governments’ expectations...

Genomics and genetics are the fastest way for a country to get into high-end biology. Why? Very straightforward. Genomics is technically modular, you can buy powerful machines, and if you have an interesting genetic question you can get it done. The technologies are applicable to plant biology, energy, environmental remediation — it’s secular when it comes to that. It’s a powerful technology that you can use for all sorts of projects to propel an emerging country to rise in the biological world.

Think about it, if you have a unique population like Iceland. You can beat out everybody. 280,000 people who had no infrastructure in biology suddenly became one of the leaders in genetics. It’s an overnight thing. Whereas you would need to build a lot of infrastructure to get a mouse facility in place and a lot of time for breeding, you can set up a bunch of sequencing machines and smart people can do informatics and you can get started right away. The beauty of that is there’s no shortage of smart people who are computationally enabled in many of these countries, and they get really turned on that their mathematical skills can have an impact in biology. Well HUGO’s role is to help these countries and these people rise to the occasion. It’s very exciting.

Japan and Australia started the biological drive in Asia. When we talk about emerging countries, we now include Korea, India, Singapore, Taiwan. South Africa is doing some interesting things, and in Latin America there are Brazil and Mexico starting significant genomics programs. So, I think we’re really talking about some powerful changes in the scientific social order. 

Will this change HUGO’s identity?

The people who need it the most right now are what I would call the emerging economies… They need a helping hand to figure out how to solve their problems, how to get the systems in place, how to frame the scientific question, how to enhance the critical mass. A big question is: What does it mean to really have genomics work for you?

… In the past, investments in health within emerging economies have primarily been viewed as a social responsibility and a resource consumer. In the last 10-15 years, health has become potentially a revenue generator because of medical tourism, because of biotech, because of pharma. And so we have seen a transition point, where governments of emerging economies began seeing their investments in biomedical sciences also as an investment in economic growth. The dual benefits have not been lost to the politicians: ultimately not only will health get better, but our economy will benefit  too because it’ll spin off into knowledge creation. HUGO cannot only assist that process, because I think you need economic development, but influence the discussion in such a way that the health component is also augmented.  Can we do that as a professional organization? Who knows, but I’d like to try.

On the rise of technology

What are the limiting factors to sequencing technology?

First of all, the $1000 genome is important because of cost considerations. It’s the same issue of connectivity. The lower the costs, the more people will use the technology, and the more useful that technology will get. The information is not just additive, it’s exponentially more valuable. The web is a classic example. At a specific saturation point, everybody has to use it. It’s not an option.

The technical limitation, really, is going to be informatics. We’re already having a hard time storing this information, let alone analyzing it. It’s only going to get worse. Now luckily, the physicists and the climatologist have worked out a lot of the issues of high performance computing. We’re still far behind them in terms of the utilization. But to catch up, we’re going to need some technologic advances to assist our computational capabilities.

Given the technical advances, do you think the days of arrays are numbered?

No. I don’t think so. But I think what it will do, is that your profit margins in arrays will fall to rock bottom — it’ll shake out. Only a few companies will be able to make them on a cost effective basis… I think the smartest thing these chip companies could have done is merge with sequencing companies. Whereas Illumina [acquired Solexa] and 454/Roche bought out Nimblegen. It’s the consolidation that mature technologies always undergo… In fact, I think it’ll be better for us in that we’ll be able to use 100 chips for the price that we use two now.

But the limiting factor is the informatics. I have a feeling we’re going to solve those issues, but it’s going to lag behind the technologies, unless there’s enough money put into it. This is where governments have to invest. Private companies aren’t going to put in millions — hundreds of millions — of dollars into supercomputing capabilities.

On Big and predictive biology

Is the Cancer Genome Atlas money well spent?

Yes! I’m biased. Having said that, could it be done better? Probably. How could it be done better? I don’t know, and we won’t know that until after the fact. Should we just sequence a bunch of tumors and in what manner, etc.? I’m much less for a rigid extraction process as opposed to a structured discovery. There is a difference. The former demands lock-step protocols and study designs for all tumors, whereas the latter takes into account the clinical nuances of each tumor type and the state of molecular/genomic knowledge. I think we’re still at a phase where we need to do a little structured discovery.

You published an excellent commentary in Cell a couple of years ago on Systems and Predictive Biology. What exactly do you mean by those terms?

The term predictive biology means to identify all the molecular components of a biological process, and to understand how the components interact so that outcome can be predicted in a de novo fashion. A lot of the stuff that we’re doing right now in transcriptional regulation in human cell lines is fundamentally to understand the complex regulatory networks at the most precise systems that I know — transcription factors and binding sites. And then to learn enough about how to manage the information and know  their dynamics so I can predict outcome given a pharmacologic challenge. In a way, this is a physiologist’s rendition of synthetic biology and I’m very keen on that.

Does that require a change in how you think about your work?

Yes, it has to. For example, gene discovery. What I used to do is do some kind of screen that would take a year or two to set up. Then painfully pull out 20 candidates. And then pray that one or two of them will be sufficiently interesting that I can spend the next ten years investigating. Now I am less concerned about the individual genes and focus on groups of genes working in concert.

For example, we worked out that the major prognostic expression profiles of breast cancers or most cancers is the proliferation index. And it’s reflected in about 1000 to 2000 genes that are consistently changing with proliferation. Any subset of those 1000 or 2000 can make a prediction of outcome in patients. Instead of focusing on prognosis only, I am asking now what are the biochemical and genetic connections for these transcriptional networks, and I can use this information to identify pathways to target in the form of new treatments. To accomplish this, we are finding both experimental and algorithmic ways to filter out, not the noise, but the concomitant roars, that come out of any process.


 This article appeared in Bio-IT World Magazine.
Subscriptions are free for qualifying individuals. 
 Apply Today.

For reprints and/or copyright permission, please contact Angela Parsons, 781.972.5467.