The Quantification of Comparative Mapping

Abstract

The estimates made on the basis of the Nadeau-Taylor (1984) model for the number of conserved segments, given incomplete comparative maps, are sometimes dismissed by the statistically naive as fundamentally flawed, due to the hypothesis of breakpoints uniformly distributed along the chromosome. (It is quite valid, however, to control for biases introduced by nonuniform distributions of known homologous gene pairs, depending on how these are discovered. See, e.g., the paper by Waddington.) No doubt there are hot spots based on local or regional sequence structures (cf. the articles by Faraut and Demongeot, Brown et al. and Demongeot et al. in Section 1), or DNA ultrastructure, gene density, isochoricity, proximity to telomere, etc. But what are the effects of these heterogeneities when the linear scale pertinent to distances between breakpoints is of the order of tens of millions of base pairs, and the time scale tens of millions of years between successive disruptions of a given chromosome? Which heterogeneities could still have discernible effects when viewed from this perspective? Perhaps those which involve a gradient from centromere to telomere, but certainly not those which refer to sequence or structural particularities recurring scores or hundreds of times along a chromosome arm. Moreover, the Nadeau-Taylor model, or the distance-free version based on gene counts per segment (which is refined and corrected in the first article in this section), are the appropriate null hypotheses against which proposals of heterogeneity must be tested. This approach is exemplified by the articles by Waddington and by Schoen in this section, which focus on the effects of chromosome length differences, as well as Sankoff et ai. (1997) investigating clustering of genes and/or breakpoints, and would be necessary to prove that there are significant numbers of segments or clusters which are protected from disruption by rearrangements.