At A Glance
- Alvis Brazma, Microarray Informatics Team Leader at the European Bioinformatics Institute.
- PhD — 1998 Moscow State University, Moscow.
- Postdoctoral — New Mexico State University.
Alvis Brazma is one of the founders of the Microarray Gene Expression Data (MGED) Soc-iety, which has been working to facilitate the sharing and standardization of microarray data. He is also the microarray informatics group leader at the European Bioinformatics Institute, and is the director of ArrayExpress, one of two public international databases for microarray gene expression data. BioArray News reached Brazma at his offices outside of Cambridge, UK, recently to discuss the MGED6 meeting held in Aix-en-Provence, France, on Sept. 3-5.
How do you think the meeting went?
By looking at the participant list, my feeling is that it was about three-fourths European participation and still we managed to get 350 participants — probably the largest [MGED meeting] so far. I have a feeling that right now Americans don’t want to travel too much, and at the same time, the participation shows the increase in interest in Europe.
Are there any applications in particular that are driving this European interest?
What has changed is that there are no longer just a few big [microarray] players. More or less now, every lab is looking at acquiring some sort of microarray technology.
Is any platform emerging as a leader?
There is really no clear leader. I think that Affymetrix is very popular but I also know, from my own collaborators, that Agilent is very popular, and many [researchers] are spotting their own. I think the market is very versatile.
How about protein arrays?
We haven’t gone into that yet: I don’t think protein arrays are ready for standardization. I would think that in the next meeting, or the one after that, [this technology] will be ready for standardization and protein arrays will be one of the sessions.
We have been writing about the FDA’s role in microarrays with an eye toward the clinical arena. Is that something that researchers in Europe are considering?
It is a thought but I don’t believe that anything has happened. I think [microarrays] are still mostly a research tool here. There was a talk by Bertrand Jordan [an advisor for the Marseille-Genopole genomics consortium], who was discussing moving towards new applications of microarrays in diagnostics, but I think it is still [about] plans instead of applications.
Let’s talk about standards: Was there any kind of consensus?
Yes, in terms of standards, particularly in terms of MAGE-ML, this was a real watershed meeting.
We talked about the MIAME experience and there were editors there from Nature and The Lancet and both of them reaffirmed the MIAME principle that microarray experimenters have to queue MIAME compliant data in one way or the other. So, the first year of MIAME enforcement from the journals really was a good experience and we are willing to continue this.
A year ago in Tokyo [MGED5], we were still talking mostly about different tools coming with MAGE-ML functionality. Now, there is a wide range of tools and lots of data encoded in MAGE-ML. So, from my own experience in ArrayExpress, we now have a well-established pipeline. Most of the Affymetrix chips are now in ArrayExpress and being submitted in MAGE-ML format. And many Agilent chip descriptions are in ArrayExpress and being submitted in MAGE-ML format. On the academic side, Stanford and TIGR are now submitting large amounts of data in MAGE-ML format. So, all of these tools are working and there are many of the microarray software vendors who already have something working, maybe not fully functional, but probably something that will be up in a few months.
What are the benefits of this?
Interoperability and data sharing — that the tools can talk to each other; that the hardware, the microarray printing and the synthesis hardware can talk to data analysis tools. And, the tools can talk to the database. Imagine if you couldn’t share genome data, and everybody would be sequencing their own genomes. Luckily that didn’t happen, and it won’t happen with microarrays. Of course, there will always be specific proprietary [information] from companies from their own compounds, but also there are reference data sets that we don’t need to replicate in many copies.
Where are the MIAME recommendations today?
My feeling is that everybody has now accepted MIAME and now there are really no more questions. The questions are now about implementing MIAME, and the answer is MAGE-ML. Of course, full MAGE-ML functionality costs some major investment money before it can yield benefits. And, even there, there is a watershed, but it is coming about. On MIAME there is no longer any dissent. There is talk about extending MIAME to chip-on-chip or the DNA location experiments. That draft is now out, and there will be comment by people who are using that in the Whitehead Institute in Rick Young’s lab. They, and the Sanger Institute, took an interest and promised to send comments. With comments from these experts, we will adjust the draft accordingly and put out the next round for discussion.
What are the opportunities for the next year?
Well, as I said, MAGE-ML is becoming a reality. What I would like to see [is] this having happened by the next year, that all the major tools, and databases have MAGE-ML import and export. Moreover, that they all have been synchronized, that all of the companies have tested it with each other, and that interoperability can really work. The MAGE-ML standard can be used in different ways, and we will still need a super agreement to make sure that we understand each other perfectly.
How would that be implemented?
On step we are taking, we are organizing a MAGE-ML file exchange jamboree for the meeting here at the European Bioinformatics Institute in December. I know that many of the major groups are coming. Also, we are working with many of them on an individual basis to see that what our ArrayExpress database can take is exactly what we can export, and vice versa.
Where do you think microarrays are in terms of development? We have written about variability. There is pressure to make the technology more reliable and hold it to much higher standards of reliability and accuracy.
I think this is happening. For the major array manufacturers, the comparisons and replicability within the same platform already is working quite well. I think the first step in understanding what different quantitation in one platform means in terms of the others — that will come pretty soon. I know that there are some initiatives to get these major manufacturers together and come to some agreement. It may also happen in this meeting I mentioned, if we have some enough time. This agreeing on some standard protocols and controls, that probably will take a little more time.
There is work being done to create standards for calibrating scanners. Will that kind of effort have an international influence?
That is all part of this technology standardization. MGED has been so far mostly involved in standardizing the data format, and moving into trying to standardizing a description of the data treatment. One of the principles of MGED has been to not to try to tell anybody how to do their experiments, but how to describe how they are doing it. But, it may well be that there comes a time that we also have to change and start issuing recommendations how to do things. So far, we haven’t really done that.
It almost sounds as if MGED has served its purpose and may have to become another organization?
No, I don’t believe that. It is just that we have to follow the main need. I think I wouldn’t make any guesses on how long we exist, but, definitely, in the near future, there is a clear need for MGED.
In Europe, how do you see the financial support for microarray research?
This is a difficult question to answer briefly. The general level of funding in Europe is considerably lower than in the US. I think that functional genomics, which includes microarrays, is having a certain boost and is doing better than science in general.
My personal major concern is not having a stable funding for the database, which is supposed to support publication data forever. So, for instance, the ArrayExpress project is about 50 percent funded from European Commission and the grant is running out in a year and a half, and we are looking for a replacement. It’s a bit too early to say if it is coming or not, and it may not be easy, and definitely it is not a done deal. So this is a problem in Europe, particularly funding for science infrastructure. It is a bigger problem than in the US, because in the US, the agencies offer renewable grants for this type of [infrastructure] activity.