DR. CARROLL: I think we can have one or two questions. We are out of time. If you could identify yourself when you ask your question, that could go into the transcript. The meeting is being recorded.

DR. YU: How about G0 expression in TALL?

DR. WILLMAN: It is really high in most TALL. So, it looks like it is going to be quite predictive of outcome in T cell ALL as well. I am sorry I didn't mention that.

MR. GAYNON: Those A,B,C and X,Y,Z, are those supervised or unsupervised?

DR. WILLMAN: Those are unsupervised. What is interesting, my point of that is, when you look at Jim's paper -- and Jim and I talk a lot at the same meetings -- most of what he published is supervised learning. It wasn't unsupervised clustering.

So, when you look at his clustering, he started with genes that are associated with karyotype and then did clustering. That is what I first showed you.

We can do that, too. When we compare our list and Jim's, they are quite similar, although we have some different genes than he does.

What we then did is back up and use VXInsight to go totally unsupervised, saying, I am not going to presume that the karyotypes are what is the primary differentiator of the data. Then you start coming up with some very novel groupings which we think are really, really interesting.

People ask me what that means a lot. I think it is interesting that maybe, while karyotypes initiate disease, by the time we are seeing leukemic children or adults, the disease has genetically progressed.

So, what we are seeing in arrays is more of a constellation of the total gene expression profile, and patients, I don't believe, necessarily should cluster uniquely to genotype, because they are going to be more reflective of many other mutations.

So, as you do the analysis of variance of these novel clusters, you start coming up with interesting commonalities and perturbed signaling pathways. It may be a different way, I think, to target patients to therapy not based purely on cytogenetics.

DR. CARROLL: So, the next speaker is going to be Tom Look from the Dana Farber, who is going to be talking about the new genetics of T cell ALL, a fish tale.

TOP