Although cancer classification has improved over the past 30 years, there has been no general approach for identifying new cancer classes (class discovery) or for assigning tumors to known classes (class prediction). Here, a generic approach to cancer classification based on gene expression monitoring by DNA microarrays is described and applied to human acute leukemias as a test case. A class discovery procedure automatically discovered the distinction between acute myeloid leukemia (AML) and acute lymphoblastic leukemia (ALL) without previous knowledge of these classes. An automatically derived class predictor was able to determine the class of new leukemia cases. The results demonstrate the feasibility of cancer classification based solely on gene expression monitoring and suggest a general strategy for discovering and predicting cancer classes for other types of cancer, independent of previous biological knowledge. The challenge of cancer treatment has been to target specific therapies to pathogenetically distinct tumor types, to maximize efficacy

We developed a model to investigate the evolution of ploidy levels in the presence of host-parasite interactions between a focal and nonfocal species. The focal species is assumed to have two loci, a ploidy modifier locus with alleles C1 and C2 and an interaction locus with alleles A and a. Thus there are four haploid gamete types in the focal species: AC1 with frequency X1, aC1 with frequency X2, AC2 with frequency X3, and aC2 with frequency X4. The modifier locus influences ploidy levels by altering the timing of meiosis; diploid zygotes of genotype CiCj have a probability, dij, of undergoing meiosis late in life, thus experiencing host-parasite selection as a diploid, versus early in life, thus experiencing selection as a haploid (Fig. 2). The nonfocal species is assumed to be a sexual diploid, having only a brief haploid stage, although results derived with a haploid nonfocal species were similar. For clarity, we assume that the A locus in the focal species interacts with a B locus in the nonfocal species, with alleles B and b. Note that Table 1 differs from this convention by referring to alleles in both species as A and a. We use a different notation here to avoid additional subscripts in the equations. The frequencies of alleles are denoted by pA, pa, pC1, and pC2 in the focal species and pB and pb in the nonfocal species, all measured at the gamete stage of the life cycle (Fig. 2). Also measured at this stage is D = (X1 X4- X2 X3), the disequilibrium between the modifier and selected loci in the focal species. Assuming an infinite population size and random mating, ignoring mutation, and following earlier work by Otto and Goldstein (1), we can write down recursions for the focal and nonfocal species after one round of selection and recombination: TX ′ = X X ( W