Skip to main content
Premium Trial:

Request an Annual Quote

Ed Liu of the Genome Institute of Singapore, on Microarrays, the Country s Multi-billion Dollar Genomics Effort, and Systems Biology


At A Glance

Ed Liu, executive director, Genome Institute of Singapore; director, Singapore Cancer Syndicate; professor of medicine, National University of Singapore.

Education: 1983-1987 - Postdoctoral fellow - Department of microbiology, University of California, San Francisco.

1980-1982 - Oncology fellowship, Stanford University

1979-1980 - Residency, Barnes Hospital, Washington University, St. Louis

1973-1978 - MD, Stanford University

1969-1973 - BS, chemistry, psychology, Stanford University.

Experience: 2001-1996 - Director, division of clinical sciences, National Cancer Institute.

1995-1996 - Professor, departments of medicine, epidemiology, biochemistry, and biophysics (UNC).

1995-1996 Professor, Departments of Medicine, Epidemiology, Biochemistry and Biophysics (UNC).

1995-1996 — Chief, Division of Medical Genetics, School of Medicine (UNC).

1992-1996 — Director, specialized program of research excellence in breast cancer (NIH Designated)

1993-1995 — Associate professor, departments of medicine and epidemiology and biochemistry, University of North Carolina, Chapel Hill

1987-1993 — assistant professor in medicine and oncology, School of Medicine University of North Carolina, Chapel Hill


After giving an hour-long sunrise session before an audience of several hundred people Monday at the American Association of Cancer Research meeting, Ed Liu stayed at the front of the cavernous auditorium in the Orange County Convention Center to take questions one-on-one.

Statisticians, clinicians, and biologists queued up for a moment with this physician and molecular biologist, who heads the Genome Institute of Singapore, and prev-iously was director of the division of clinical sciences at the US National Cancer Institute, and before that, a professor at the University of North Carolina, Chapel Hill. In 2001, he joined the Singapore Institute, which was founded in June 2000. Last year, the institute was among the first groups to sequence the SARS virus, using samples from among the first patients, or index cases, in the Asian city-state to derive a sequence in a period just over two weeks.

Liu spoke to BioArray News this week at AACR about microarrays and how his institute uses the technology in developing its approach to systems biology.

How did you first get involved with microarrays?

When I first started out, when I was at NCI, as scientific director for the division of clinical sciences as an extramural investigator, I was asked to sit on some boards, one of which was, in effect, to examine new biomarker platforms. This was at the end of 1995. I joined in 1996, and once I joined NCI, I became an intramural representative on that board and that is where I got acquainted with all that technology, and I got to know the Affy guys, when they were first starting out. I also knew Pat Brown; we were postdocs together at Stanford.

We waited until 1997 to do our own arrays because Affy and Synteni kept saying ‘Don’t bother making your own, we’ll do it.’ It never happened because they were busy trying to sell to pharma. I got pretty frustrated about that and decided that we would start our own intramural program.

What caused your frustration; what did you see in the technology?

This was the most remarkable thing that I had seen in a long time and I wanted access to it. In those days, we were looking at five or six biomarkers in a [pool] of 500. In those days, that was a large number. Then the opportunity came to look at a 1,000 at one point. So, we got started with the help of Jeff Trent, who was at NHGRI and with Pat Brown, who was always extremely generous. We made our own printer, our own reader. We constructed our own informatics system within the intramural program — which we now use at the Genome Institute. We got to understand the technology very well, and the limitations. We moved from cDNA, to purchase PCR products, because even for us, to manage the quality control of 10,000 clones was just too much. In late 1999, we started to work with Compugen, which, at that time, was a very small company. We gave them a bench with us at the NIH and we said, give us access to your oligos and we will help you figure out how to use those oligos in a spotted array. We basically did all of the original quality control. We use their product now as our fundamental provider.

When did you go to Singapore?

In March 2001. It was originally called Singapore Genome Program, and I was on the board. The government had asked a small number of people to do that, and the next thing I knew they asked me if I wanted a job. It was a real unique opportunity because of the amount of money they were going to put into it — literally a sweeping change of the infrastructure. We are talking about changing secondary school education, changing the patent laws, everything was going to be changed. A lot of people have this impression that in Asia they will cut corners, but they don’t do this. What they want to do is move things fast, but ethically. It has been a pleasure to work in Singapore because of that. I don’t feel like we are doing anything that is out of the realm of honesty.

Did the mission of the institute change because of last year’s SARS outbreak?

Actually, SARS was an exercise of the mission statement we had made before that. I wanted to have a holistic structure that would take basic biomedical science, using engineering and high throughput means, and put cell biology in the middle of it — to give it a certain intellectual substance, and then address certain human problems pertinent to health, economy, and society and not being specific in defining it. What happened with SARS simply allowed us to exercise our vision. We were constructed in a way that would have minimum boundaries between technology and biology — have a group of young dynamic people who would see the need for our intervention and go for it.

Did you feel you performed well in that crisis?

I think we surprised people. What it did in Singapore was also [to] completely cement our relationship with the senior leadership of the country. When you embark on a three to four billion-dollar effort, in US dollars, in a short period of time, for a small country that has 4 million people, like that, a few eyebrows are raised.

Now, we have high quality computation people all around us, all eager to do biology. We have sufficient resources, and I don’t mean exorbitant resources, but sustained, and reasonable, and HR policies to recruit and maintain them. So, the only limiting factors are the culture of the place. Of course, if you could hire a few Nobel laureates . . . but that’s not going to happen so you have to make use of your talent, and whatever talent you can attract. We have a staff of 190, and our administration is lean and mean, with about 12 percent of that. Everybody else is somewhat devoted to science.

How are you using microarrays?

Our array program, as are all of our technology programs, [is] completely integrated into the biology questions we ask. We collectively talk about where we want to go. Our whole system is encouraged to have people working together to have an impact on the scientific questions asked. All of the biologists in the institute use arrays in one way or another. Our platform is now split into two formats. One is our homegrown oligo-spotted arrays — human, mouse, zebrafish and Schizosaccharomyces pombe. They were constructed because of the possibilities within scientific linkages that could be done within Singapore, in corsortia, or within the institute itself. We use our homegrown technology for that, because our cost structure is quite low. I can’t give you the exact number, but it is considerably lower than anything we have bought.

Is it lower than $100 per chip?


Significantly lower?

Well, it depends on how you count. I’m not trying to hedge but it is hard to give an exact cost. We don’t pay rent.

Affy and NimbleGen are our major vendors right now. We are doing all of our human tumor stuff on the Affy. On the NimbleGen platform, we are doing a lot of our exploratory work, including infectious diseases. And, then our spotted arrays, we use for our cell biology. You know, to do a good cell biology experiment, you have to do 50 at a pop. You do different replicates, different time points. For example, the scientist who is doing our pombe arrays did a time course experiment with 10-minute time intervals, in duplicate, over something like 210 minutes to get at two and a half cycles to get the oscillations, and it was at that range that you could begin to see the resolution that you really can’t pick up by any other means.

We can only do that [volume] because we are not a single lab. You need volume to keep the cost down. If you are a single lab trying to do this, you will break the bank, even if you make your own. We think we have a real competitive advantage. We have great technologists, highly flexible structure, some excellent biologists asking some pretty interesting questions in cancer, stem cell biology, pharmacology, and host-pathogen interaction.

How does your vision of integrated biology compare to that espoused by Lee Hood.

They are pretty much the same. He and I are on the same circuit because we are proselytizing the same thing. Lee has a greater history in what some people call pure systems biology where you are really trying to quantify a single biological process and then model it mathematically. Most of the stuff that we do is not amenable to that kind of computation at this point but there are qualitative models that we are trying to put in place. They are more discovery. You have to pick your battles, you have to choose a biological system that is amenable to that type of analysis. What Lee did with Eric Davidson, that was a brilliant stroke. The system that they used, the sea urchin development, is beautiful. It is forward decision making. The whole concept of that is that you are not talking about homeostatis. You are setting something in motion, the domino [effect] takes place, where you don’t have feedback checkpoints, negative regulators, what have you. Your modeling process is a lot easier where you don’t have to deal with negative feedback loops, and everything else. They even readily admit that.

You have implemented some common technologies in the institute. How is that working?

People are recognizing the value of this and wanting to access it. Even five months ago, we couldn’t say that. But we aren’t the only ones. Several pharmaceutical companies are moving in the same direction, [integrating] a common database, a common array platform, a common informatics platform, with multiple biological groups using the same platform so that the informatics guys can mine that data over time.

Are you putting all of your data eggs in one basket, in terms of your platform selection?

Well, we are really not because our cell biologists are using the spotted arrays, and that is really a cost issue. We now are relatively comfortable with our platform, so that a good array, any good array of any platform, as long as it is good, the raw data is very comparable. Now the clinical stuff, that is another challenge. My concern is that the current platforms for microarrays are not really the optimum platform for clinical expression profiling. It is sensitive enough? The dynamic range is not really good and it has process problems. The performance has to improve, and the cost has to go down. I just don’t know if we will hit a limit beyond which no two-dimensional system, using fluorescence, is going to be able to perform as you want it to. We are exploring mass spectrometry, and different expression detection systems. And, we are getting to be known well enough among platform companies that a number of them are coming to us with some of their more prototypic devices. Sequenom has an expression platform, ABI has one, Illumina has an interesting one as well. All of these guys are coming out with really interesting products.

Have you looked at the ABI Expression Analysis platform?

We are actually beginning to look at that now. We are talking with them. The issue with ABI for me is follow-through. They start a lot of things, but they don’t take a lot of them to completion, and they always come back to their core of sequencing. It’s just a matter of how hungry they are. We are impressed some new technologies coming out of England — Solexa.

What is your take on the Asia in a microarray-based diagnostic context?

Historically, the diagnostics market has been unregulated in Asia. I always sense that the bioentrepreneurs in Asia are very willing to start with diagnostics. They are used to skimming small margins out of technologies. They aren’t phased by it. Trying to do a diagnostics company in the US is almost impossible. Nobody is going to fund it, they will say, ‘Those margins are too small.’ Taiwan, for example, is not afraid of small margins. They just want to be a player because they know that at some point, they can beat everybody else by dropping the costs. It’s a mindset. But, is there a market for it? That’s a different story. Asian countries have always had a tendency to be less socialistic about their healthcare systems than Europeans. A lot of out-of-pocket costs are paid. What does that mean for diagnostics platforms, I don’t know. But if the unit cost of a diagnostic, which usually is lower than a therapeutic, is coupled with a less-regulated environment, you can imagine that the patients themselves will be calling the shots, asking for a particular diagnostic as you would a pregnancy test. In the US, you have this weird situation where only the doctor can prescribe a test. That is why I think the diagnostic market in Asia has very reasonable potential because of the social and economic structure.

How do you think microarrays can be improved?

In terms of clinical utility, the miniaturization and the amplification aspects [of microarrays] will be a key, but the other one is the utility of this kind of platform in paraffin-embedded sections, which has already been described. That is going to accelerate discovery immensely because you can go retrospectively into large tissue banks and actually get that done. You extract RNA off a slice of paraffin-embedded tissue and use that with multiplex RNA diagnostics. On a clinical basis, take a tissue and slice it up and I can do a 100-gene expression off that and that would be quite helpful.


The Scan

Study Points to Tuberculosis Protection by Gaucher Disease Mutation

A mutation linked to Gaucher disease in the Ashkenazi Jewish population appears to boost Mycobacterium tuberculosis resistance in a zebrafish model of the lysosomal storage condition, a new PNAS study finds.

SpliceVault Portal Provides Look at RNA Splicing Changes Linked to Genetic Variants

The portal, described in Nature Genetics, houses variant-related messenger RNA splicing insights drawn from RNA sequencing data in nearly 335,700 samples — a set known as the 300K-RNA resource.

Automated Sequencing Pipeline Appears to Allow Rapid SARS-CoV-2 Lineage Detection in Nevada Study

Researchers in the Journal of Molecular Diagnostics describe and assess a Clear Labs Dx automated workflow, sequencing, and bioinformatic analysis method for quickly identifying SARS-CoV-2 lineages.

UK Team Presents Genetic, Epigenetic Sequencing Method

Using enzymatic DNA preparation steps, researchers in Nature Biotechnology develop a strategy for sequencing DNA, along with 5-methylcytosine and 5-hydroxymethylcytosine, on existing sequencers.