AT A GLANCE
Senior Scientist, director of OmniArray research program, Singapore National Cancer Center.
Research focuses on gene predictors for breast and stomach cancer.
Received his MD PhD from Stanford 2000. Worked on C. elegans microarrays with Stuart Kim.
QThis Saturday, you are holding a one-day Singapore Microarray Meeting at the National Cancer Center there. Is this the first of its kind?
AThere have been meetings where institutions have invited researchers from the US, the UK, and Europe to give talks on microarrays. But this is the first meeting in Singapore that has brought together researchers from a wide variety of different areas, with a specific focus on the microarray biochip groups working in our local research community. The purpose is to get them together to share a little bit about what they are doing, the problems and challenges they are facing, Certain collaborations might also arise from the meeting.
QSingapore recently earmarked an astounding $4 to $5 billion for life sciences. Has this increasing interest in microarrays arisen out of that commitment?
AOver the past couple of years the government has said, “our electronic and IT industries seem to be chugging along. Let’s concentrate on a new pillar to build the economy.” The next obvious thing is bioscience, and they have earmarked about $1 billion to develop that industry. It isn’t just research money, it’s money to get venture capital in, to develop the institutions that are going to support that type of work.
QHow did you get into doing microarrays at the Singapore National Cancer Center?
AI got my MD PhD from Stanford in 2000, and basically if you come from Stanford and were doing biological research during that time, you were doing microarrays. Singapore is my home country, so it was not a super-sophisticated decision to try to apply some of this technology in a local setting.
QHow do you use microarrays at the Cancer Center?
AThe National Cancer Center is the major public sector cancer health care institution in this country. Seventy to 80 percent of all public sector cancer patients come through this facility. The people who started the cancer center here four or five years ago established the banking of tissue as a routine procedure. The tumors are flash frozen straight from the operating room and stored in liquid nitrogen. The tissue is good quality, and you can do RT-PCR and arrays with it. The other thing that’s interesting about this population is that certain diseases differ from those in the West. One example is nasopharyngeal carcinoma, which is not that well represented in the West but is a big problem in Singapore and China. We are trying to apply some of these genomic technologies to understand these diseases.
QWhat kinds of microarrays do you use to do your research?
AWell, if you come from [Stanford] you have to make your own. But we have also been using Affymetrix arrays for the past several months. You can get very good results from the Affymetrix arrays so long as you can afford them. For much smaller projects where the sample provides only enough for one or two arrays at the very most, this is a good choice to put on Affymetrix arrays.
QWith these tissue samples, do you generally have problems getting enough RNA to put on a chip?
AIt depends on the project. With liver, we can go with total RNA. With breast, we amplify the RNA. With [laser capture microdissected] tumor cells, where you are down to the nanogram levels, we do multiple rounds of amplification. There are amplification biases, but that’s why we use RT-PCR based on the original RNA sample to validate all of our results.
QThere has been a great deal of discussion about which microarray analysis methods to use. What are your opinions on the issue?
AI think right now nobody has the answer as far as what is the most perfect way, but it’s clear that current algorithms are sufficient to give you some biological insight into what’s going on. We use all sorts of algorithms, hierarchical and agglomerative clustering, PCA, support vector machines, neural networks. We create everything in house, and write all our code, mainly because we need to be able to muck around with our data. Also, if you look closely at a lot of these algorithms, in the packages, you can get them for free. For example if you call [Michael Eisen] he will send you the clustering package free of charge.
QHow many people do you have doing bioinformatics in your team?
AWe have about five people spread out among different aspects. Some people are doing the databasing, others are doing actual implementation of the infrastructure, and others are doing algorithms.
QAre there any particular challenges or advantages to doing genomic research in Asia?
AThe advantages are that it’s nice to come into a situation where you can build from the ground up. There are no legacy issues with technology. Second, the biological questions you have are different from those in the West, and in the Cancer Center, it is easy to get the numbers [of samples] you need to do large-scale profiling studies. A main challenge, though, when you bring in a new technology, is that there are not many people that are trained to use it. You have to do a lot of things by yourself. That means moving from the informatics to the wet bench, and talking to the clinicians. Until everything is up to speed, you have to do more leg work. The other challenge, which is eased to a certain extent with the Internet, is that we are a bit removed from the cutting edge of the field in the States. Also, you can’t assume that you are going to get something that you have ordered right away. With certain distributors in the US, where you are used to sending an e-mail and getting the product shipped in two to three days, it may take six weeks to come. In your work you have to take a much more long term view of what your needs are, and you have to be a bit more organized. Lastly, a lot of the big companies that offer good microarray instruments have only recently tried to find distribution representatives in Singapore.
QDid you order an arrayer, or did you make one yourself because of this difficulty?
AWe did buy an arrayer. But part of the choice in instruments here is the support available. Something that may be less good on one front, if it comes with very good tech support, you have to go with that.
QWhat do you think is the biggest limitation or obstacle in microarray-related research and what is your wish list for key technology improvements?
AThe biggest obstacle is how you compare data from one group to data from another. It’s an issue even with the Affymetrix arrays, and an even larger issue for the two-color arrays. If we don’t solve this problem each dataset is limited to [relevance] within the group in which it is created. What you would like to have is a GenBank for microarray data. And that’s a big issue in getting people to willingly contribute that data.
As far as improvements, I think they are already coming. DNA arrays are moving away from being research tools to being service tools. With the increase in core facilities, microarray work in the next five or six years is going to be like DNA sequencing. The challenge right now is to start using the data you generate. Also, there is room for improvement in protein chips.