HETEROGENEITY OF MICROSATELLITE MUTATIONS WITHIN AND BETWEEN LOCI, AND IMPLICATIONS FOR HUMAN DEMOGRAPHIC HISTORIES

Microsatellites have been widely used to reconstruct human evolution. However, the efficient use of these markets relies on information rega rding the process producing the observed variation. Here, we present a novel approach to the locus-by-locus characterization of this process . By analyzing somatic mutations in cancer patients, we estimated the distributions of mutation size for each of 20 loci. The same loci were then typed in three ethnically diverse population samples. The genera lized stepwise mutation model was used to test the predicted relations hip between population and mutation parameters under two demographic s cenarios: constant population size and rapid expansion. The agreement between the observed and expected relationship between population and mutation parameters, even when the latter are estimated in cancer pati ents, confirms that somatic mutations may be useful for investigating the process underlying population variation. Estimated distributions o f mutation size differ substantially amongst loci, and mutations of mo re than one repeat unit are common. A new statistic, the normalized po pulation variance, is introduced for multilocus estimation of demograp hic parameters, and for testing demographic scenarios. The observed po pulation variation is not consistent with a constant population size. Time estimates of the putative population expansion are in agreement w ith those obtained by other methods.