Tech joins battle against HIV

Armed with medical and doctorate degrees from Stanford University, Heckerman has virtually come full circle in the last 15 years. As lead researcher of Microsoft's Machine Learning and Applied Statistics Group, Heckerman has seen technology he helped develop serve as analytical tools in HIV research.

And earlier this month, Microsoft made the technology available to the research community, as open-source code for four analytical software tools to develop a vaccine for the disease.

Heckerman, who previously served as a university professor in computer science and biomedical research, joined Microsoft in 1992 at the request of then Chief Technology Officer Nathan Myhrvold. Heckerman recently talked with CNET News.com about his vaccine work at Microsoft.

Q: Given your background, how did you end up at Microsoft?
Heckerman: I initially started with an MD degree and quickly got interested in things like artificial intelligence, probability statistics, that sort of thing. I started developing tools that would help physicians do diagnosis and, when I got out of school, my intention was to do biomedical research.

I was doing that when Microsoft came a-calling. They were interested in the Ph.D. work that I had done, the diagnostic work. They saw that it could be used not only to help physicians do diagnoses on people, but it could help computers diagnose themselves. So they convinced me to come. I came kind of as an experiment, just to see, and I thought, you know, I'd be there at most maybe five years to see what it was like.

The applications doing diagnosis worked out very well, and then I got interested in all sorts of other things.

What were these other things?
Heckerman: For the first 10 years I was (at Microsoft), I worked on data-mining tools for SQL Server. I also built spam filters and worked on some early health systems for Office. Then, maybe over 3 or 4 years ago, I began to see that a lot of the technology that I had developed at Microsoft over these 10 years was also very useful for the original things I was interested in, namely the biomedical applications.

Microsoft Research is very good about giving us a lot of freedom to innovate, and they said, "Sure, go ahead and try that out," and very quickly, I found some applications for this technology that I developed here at Microsoft for the HIV vaccine problem. That led to a couple of early successes, which connected me with a lot of people in the HIV research community. Then, for the last three or four years, I've been working almost exclusively with them to help create this HIV vaccine.

How much support did you receive from Microsoft when you began this project several years ago?
Heckerman: Well, initially, it was just myself. Then there were a couple of other researchers, a programmer, a couple of interns to work on this problem. In some of the cases, the work that we did needed to be experimentally verified, and so Microsoft paid for some of those experiments. Now the actual cash outlay is in the hundreds of thousands of dollars, but if you include the cost of us, the researchers, it's well over $1 million.

And before you developed analytical tools for building HIV vaccines, what was there before?
Heckerman: So, prior to us coming on the scene, people were using mostly straightforward statistical analyses--the kind you'd find in Excel, for example.

I can't say we're the only ones doing advanced things in this area. Certainly, there is a lot of good work out there, but what we did was, we looked at the problems that biologists were trying to solve, and we said, "Well, there are some very high-tech things we've done here that can help them better solve those problems," and we developed tools around those ideas.

Since developing those four analytical tools in 2005, what breakthroughs have happened in AIDS research?
Heckerman: One of the things that we're doing is trying to evaluate how effective our immune systems are at fighting HIV. There are actually two arms of what's called the adaptive immune system, the part of our immune system that can adapt to a particular bug or pathogen that might infect us. There is the humoral arm, which makes antibodies, and there is the cellular arm, which uses essentially white blood cells to attack our own cells that are already infected by a virus, killing our own cells so that the virus or some other disease doesn't spread.

Since HIV was discovered about 20 years ago, most of the research community was focused on the humoral arm of the immune system--how to make a vaccine in terms of antibodies--and there hasn't been a great deal of success in that area. And more recently, researchers have been turning to the cellular arm of the immune system. There has been this debate about how effective the cellular arm of immune system is against HIV.

So our tools help answer that question, and the way we're answering that question is, we're looking at how HIV mutates. HIV is a virus that mutates very rapidly; we're looking at how those mutations in HIV correlate with our cellular immune system.

It turns out, we all have different immune systems, and we can quantify that in the lab. You can measure what your immune system is, and then we can take blood from you if you're infected with HIV, sequence that HIV, and then look for correlations between what kind of immune system you can have and what kind of mutations happen in HIV.

If there is a strong correlation, then that means, or it certainly strongly suggests, that our immune system does have a profound effect on HIV, which then suggests that we can build a vaccine to fight HIV.

What is your greatest hope from having these four tools out there? Do you think that in five years from now, maybe a vaccine will be developed to eliminate AIDS?
Heckerman: The development of a vaccine for HIV is clearly not an easy thing. People have been working on it for 20 years. My hope is that we can contribute somehow, and it looks like we are doing that.

In terms of specific details, now that we know what the immune system does to fight HIV effectively, the next question is, where are the signals on HIV that we can exploit for a vaccine?

The immune system doesn't look at HIV as a whole. Rather, it looks at it in little bits and pieces. We need to find those bits and pieces, so we can put them in a vaccine and train your immune system ahead of time, before you get infected.

There is a search now. It's kind of a needle-in-a-haystack search for those signals within HIV that our immune system uses to attack it. Our tools are useful for finding those bits and pieces of HIV.

As you worked on these tools, did you think you'd ever go back to the pure biomedical research you were getting under way?
Heckerman: The work we're doing is so exciting that I don't see any stopping in the near future. When I came to Microsoft, it was mostly an experiment. I thought maybe I'd be there at most five years, but with all the freedom that they've given me, I'm ending up doing what I was going to do anyway, namely a mix of computer science and biomedicine.