NEW YORK, April 19 - About two years ago, confronted with a luffing cell-cycle project, Jeff Voss, a molecular biologist at Abbott Laboratories, approached the company's CEO with an idea.
"I said, 'You know, nobody's really doing genomics here.' And he said, 'It sounds like a good idea.' But he didn't give me any resources to do it," said Voss, who today is genomics and bioinformatics team leader of the Abbott Bioresearch Center, in Worcester, Mass. "'Fine,' I thought. 'I'll do it on my own.'"
So Voss recruited a couple of undergrads from a local college and began running filter arrays. "The data we got from that platform were enough ... to get justification to move into the Affymetrix platform," he recalled in a recent interview with GenomeWeb.
During the first stage of the team's research, "we just went along and took what organs we could get from existing investigative tox experiments [going on in the company]. We'd just go down and beg for liver slices," Voss said.
Soon, Voss' lab grew. After a year of proving the Affy platform can help his nascent team collect convincing data, his group swelled to eight FTEs, including the one post-doc.
The lab will soon undergo another change. About two weeks ago, Robert Kamen, the executive who headed the ABC site for virtually its entire existence, retired. Voss and Kamen's replacement, Alejandro Arrufo, are planning the future of Abbott's genomics research, especially the toxicogenomics program that has become vital to the pharma giant's drug discovery.
GenomeWeb: Briefly, what kinds of research does your lab perform?
Jeff Voss: Primarily, we run Affy gene chips in various capacity. We also do some work making subtracted-normalized cDNA libraries using animal models, and we spend a small amount of our time performing database searchers.
We use Affy chips in three ways: We have invested a lot of resources in doing toxicogenomics, we perform compound profiling studies, and we use expression profiling to look for surrogate markers for effects of different cytokines to learn the exact mechanism by which these cytokines exert their influence on different types of cells.
GW: What tools or technologies, beside the Affy chips, does your lab use?
JV: We have a Roche LifeCycler and we have an ABI 6700 high-throughput QPCR. Our throughput on QPCR doubles every month, which was the main reason to make the switch from the Roche machine to the ABI platform.
We also use the Rosetta Resolver and we use Spotfire [technology] pretty heavily. And, actually, in our site I'd say our use of Spotfire is pretty higher than [our use of the] Resolver, but in Chicago (where ABC's sister facility is based) it's probably the opposite.
One of our biggest challenges is reporting data back to the product teams [throughout Abbott]; there's a lot of data there. And Spotfire is really useful for visualizing and condensing that into a pill that a lot of people can swallow. But it still goes by pretty fast.
When we stand up at a project-team meeting and present the data, the visualization tools are not the sort of thing that people are used to seeing. So we've created an Internet site where we will post these presentations, and we have put them in a form that you can pull up presentations that relate either to compounds or to different model systems or different indications.
For this, we take a set of visualizations created in Spotfire, put them into PowerPoint, and convert them back into HTML, which we put up on the web.
GW: What a cool idea. Would you ever commercialize this technology? It sounds as though it may help other pharma companies faced with similar challenges.
JV: We would never do that. But I think the concept is something you can probably shrink-wrap and sell. I'm not the only one who shares this sentiment. You can do a lot of stuff with expression profiling, but if you're a project-team leader it's not on your critical path. That's one of my big jobs right now is to develop applications to put on the critical path. I can't do that unless people are looking at the data. So I have to make it as accessible as I can.
Also, making it into a web tool for the project-team members allows me to monitor how often people are looking at it.
GW: Are you generally satisfied with the tools you use?
JV: Oh, absolutely. Spotfire's technology? I can't say enough about it. It's really economical, and the value of it is as high if not higher than anything out there.
Affy is pretty pro-active, and I'll get a lot of e-mails and phone calls from them asking 'What can we do?'
ABI, though, is a little different. We've spent a fair bit of money on ABI products, and their service has not been that crack. We had this one situation when we bought the 6700--that's about $100,000, so it wasn't cheap. We took early delivery of it in December, and it wasn't actually functioning for three months. They brought it in, it sat here, they couldn't get a rep out here to set it up, and when they did get it set up there was a major mechanical failure. And then they needed to send somebody out to a training session, which took another two or three weeks. I wasn't very happy with it. They wound up having to give us some reagent credits.
If I was a rep who sold such an expensive tool, I'd be at the site opening the crate myself.
Having said that, ABI does have an instrument that I'm trying to get some support for. It's a robotic station that makes RNA as well as set up reactions. You can sit it next to the high-throughput QPCR machine.
The way I see our QPCR usage going, having that automated capability would be really nice. It's just that the size of my group would kind of prevent it us from getting funding for that tool alone. But the fact that it sets up sequencing reactions brings it into a niche where I can justify it now because we can use it for more than one thing.
That's probably the next thing we'll buy from [ABI] in spite of the fact that we weren't that happy with them [in the past].
GW: If you can change anything about the tools you use, what would that be?
JV: You know, the price of the Affy chips is just unbelievable. It's really something usurious. We run a lot of chips, and even if you look at the discount rate we get ... highway robbery comes to mind.
When we were calculating our budgets we were expecting the prices to come down because there was a lot of competition, but that's all pretty much gone away. I mean, Corning doesn't make chips now, and I thought they were going to be Affy's biggest competition. Agilent was going to start printing oligo chips using the Rosetta technology. There's been no competitive influences to bring down the price of chips.
Affy has also done something that I've found a little intimidating: They've come out with a new human chip and they're discontinuing the old human chip. And they recognize that 'Well, people are used to paying a certain price per chip,' so Affy took all the genes and they divided them on two chips, whereas before they had a couple of chips and they had basically all the good genes on one chip and all the stuff you weren't that interested in on another chip.
Now they split down the middle, so in order to get good coverage you have to buy both chips. So now the list price is $2,000 for a pair of human chips to get coverage, whereas before it cost about $1,000.
Having said that, this more expensive pair of chips is actually better. In total there's more genes on there, and Affy's laundered out the ones that didn't work before. It's a lot denser.
GW: Is it worth a doubling in cost?
JV: I don't know, because we haven't used it yet. But we have one project that we'll apply it to within a couple of weeks. But they're so expensive. If you think about it, when we do a tox experiment, we're talking 10 or 20 chips. And at $1,200 for the pair [with our discount], that's a lot of money. And unless you get something really useful out of this it raises a lot of eyebrows.
What Affy will do, though, is custom-make for you a smaller human chip. So I think what we might end up doing is having a bunch of those custom made and use them for most of our human stuff. And if an experiment looks really good then we may use those same probes on the more expensive chips.
GW: Has Abbott tweaked your budget to meet those price increases, either for the gene chips or for traditional creeping costs?
JV: We're kind of in a honeymoon period. We got an allocation to get this research going, and in a year or so we'll probably get evaluated. Frankly, I think we're doing really well. We've been able to say many things about the compounds we've made.
GW: If Abbott were to give you carte blanche to go shopping for a genomics tool--whether it exists or not--what would you buy or ask a tool shop to build?
JV: Two things come to mind. One is that I need some sort of informatics tool that lets me annotate genes more efficiently. What we spend the most time doing is looking at a transcript and trying to decide what that gene does. And of the many things that the gene product might do, which one is relevant to the change? We try to put this information together in a way that will link genes together.
So if we can establish through reading articles some sort of epistatic relationship between the genes or some sort of sub-unit or physical interaction in relationship between the gene products, then it enables us to pinpoint a pathway that the compounds might be affecting.
And that's hard. I mean, right now we spend a lot of time just going through PubMed, you know? There are some other ways to do it--to support the annotation of what we're doing--but they're in their infancy. There's PubGene, which measures the number of times two terms or co-genes are co-cited in an abstract in PubMed. So you can use those search terms to link genes together. These are the things we'd be willing to pay for.
We also need to find a way to describe genes in short text terms that can be clustered. That's not really a product, but if there was a way of systematically clustering genes based on aspects of their function ... that would be useful to us.
Actually, Proteome has some of this stuff, and we considered licensing some of this stuff a year ago, but we didn't have the money. But right now we're actively in the process of reviewing some of these things and we're going to make some decisions about what we're going to license.