Skip to main content

Paul Van Hummelen, Research Manager at VIB Microarray Facility



PhD in cytogenetics from Flemish University of Brussels, 1997.

Postdoctoral fellowship at Lawrence Livermore National Labs.

In April 1999, opened the microarray facility serving the Flanders Institute for Biotechnology (VIB) in Leuven, Belgium,

Interests include: Scuba diving and rock climbing.

QHow did you go about setting up a core microarray facility from scratch?

AThe facility opened on the first of April 1999, which is Fools day in Belgium. There were two issues. The first was time. If you’re going to invest the money you have to be on top of the technology and be ahead of other people. So we wanted to go for a program that had everything in it, and that was Affymetrix or Amersham. Price-wise and service-wise we went for Amersham. They included an arrayer, a scanner, and a protocol. They promised a user group where you could exchange information. They had their own little database to track all the samples, which is very helpful as a startup. But the database is very limited and now we’re building our own one.

QDo you still plan to continue making your own arrays, with all of the prefabricated arrays out there now?

AIt depends. We have our own clone sets now, but if there are interesting arrays coming out dedicated for certain issues or certain cancers, and some people are interested in using them, then we will. If it’s cheaper for us to buy them than make them ourselves we will do it, if [the array] meets our standards.

QHow is the facility set up within the university?

AThe facility is only for Flanders Institute for Biotechnology (VIB) researchers. We do everything for them and they pay the cost of making the array, including the personnel and the machines, because we want to be a self-sustaining unit. We have five people. The institute is about 850 people, and there are 300 researchers. We did something like 50 arrays last month.

QDo researchers come into the lab and watch you do their microarrays, or are they hands off?

AThere are researchers who want to see everything, and follow everything. We allow them to do that. First we have several discussions with them about the experiment. We see what data they want to get out of it, and how to design their experiment. Then they send the sample and we do everything. We have data analysis software in the lab. We don’t want to be a pigeon houseóas in, the postdocs come in, they do the work, then they go out. Because otherwise our quality tends to drop.

QHave you developed any protocols in-house?

AI spot in 50 percent DMSO to prevent evaporation. I cross-link the arrays immediately after spotting, and then I store them somewhere dry. Before I start the hybridization, I put the slides back in a humidity chamber, at, say, 90 percent humidity. Then after ten or twenty minutes I take them out, let them dry again and cross-link them. Suddenly the signal is boosted. We did this with the first arrays we ran. It produced a dramatic effect.

QWhat is the biggest challenge you face in working with microarrays?

AThe most important variable is the quality of the RNA. If you have good quality RNA most of the things will work. We have an Agilent Bioanalyzer to look at RNA. But what researchers usually underestimate is the quantity and quality they give me.

So now we do amplification, using linear amplification based on the Eberwine protocol. We tested it out and it does not bias the results. Everyone should test out the protocol they use, because you may get a bias.

QWhat’s the biggest mistake that researchers do make?

AMost researchers try a kit, see that they have very nice pictures, and then think it works. But this is not true. You can have a lot of bias in that picture. Poly (A) effects are very important. You have to block for Poly (A) somehow. Or it will have cross-hybridization.

To test for cross-hybridization, do a basic experiment that you repeat from the beginning and always take with you through the years. I make a new array, and immediately hybridize it with the vector primer, to see what the quality of the spots is. Second, I do a mouse spleen-spleen and heart-spleen. Because I know what the results are, and if I get the same results as I have before, there’s [consistency].

Or you can take 2 mcg spleen against one mcg spleen, and then you mix it with something else that is one-one. If you plot intensity vs. intensity, you should get a two-one line, and the other genes should be on the one-one line. But if you have cross-hybridization with some of your probes, everything is on the one-one line because it’s all the same. This is very important; if you are not careful you are looking at totally wrong results.

QDetermining the lower threshold for significant changes in expression levels seems to be a problem in analysis of microarray data. How do you deal with this?

AWe haven’t found a good solution yet, but you have to put a threshold on your data somewhere. I put the threshold at somewhere around two standard deviations above background. It’s very hard to explain that to researchers, because there are 5,000 genes [on the chip] and they want to know about everything. They say, ëwhat if you have a transcription factor that has very low expression levels? It’s going to be below that threshold.’ I say ëyes, but its not reliable [data] so what will you learn?’

QSo you should always err on the side of a higher threshold?

AThe design of the experiment comes into play. If you have a time point experiment or you are looking at knockout vs. wild type and you can do in vivo, vs. cell lines, induced or not induced, you can do a diagram and see which genes are common to all the different systems. It’s just, ëare they consistently over- expressed in different experiments?’ You can go lower, to a 1.5- or 1.3-fold change. If they are still consistent, you can believe it. But black and white experiments are the worst experiments you can do. [Comparing] something against something else. You have to filter out the cause and the consequence.

QWhat would be your wish list if you could have any tool imaginable to improve your microarray experiments?

AI want a high -automated probe-making device. We spend a lot of time doing every probe sample tube by tube because we’re afraid that if we do it in high-throughput we lower our quality. But most importantly, I would like to have a whole department of bioinformatics right in my lab.

It’s very hard to communicate with bioinformaticists even if they are in the same building but a floor down, because they have their own projects they are working on. They calculate ëI spend 15 percent of my time on the microarray business.’ But you have to spend 100 percent of your time on it.

The Scan

Gap in COVAX Doses

BBC News reports that COVAX is experiencing a vaccine shortfall, as the Serum Institute of India has paused exports.

Sanofi, GSK Report Promising Results

The Wall Street Journal reports that the candidate SARS-CoV-2 vaccine from Sanofi and GlaxoSmithKline has had encouraging early results.

Influence of Luck

The New York Times examines how the US avoided variant-fueled increases in COVID-19 cases.

PLOS Papers on Retina GWAS, Hantaan Virus, COVID-19 Phenome-Wide Association Study

In PLOS this week: genome-wide association study of retinal morphology, analysis of hantaan virus found in a mouse, and more.