Prediction and analysis of spatial order in haploid chromosome complements

Bennett has proposed a model that predicts a mean ordered arrangement of all the chromosomes in a simple haploid genome, based on associations of pairs of most similar long, and pairs of most similar short, chromosome arms. The model orders a complete simple haploid genome so that each chromosome is associated with two constant neighbours. This paper describes a test of the model with two types of data obtained from the same reconstructed serially sectioned somatic metaphases examined in the electron microscope. First, chromosome arm volumes were estimated and used to identify the chromosomes and to predict their mean spatial order. Secondly, centromere positions in three dimensions were established. In the species and hybrids used, all with 14 chromosomes, there are so many ways of positioning the chromosomes within haploid sets that a computer-aided analysis was developed. With use of only centromere identities and positions, the programs generated all possible orders of centromeres in haploid sets (where each centromere has two neighbours) and computed the sum of distances between centromeres for each order within a cell. Orders were ranked in ascending sequence of sums of distances. Orders that ranked highest were taken as ‘best’. After results for replicate cells had been pooled, orders were ranked from best to worst. A test of the predicted order was then made by finding its position on this summary. In all the four grasses examined, the predicted order was among the 5% of orders judged best by the analysis. To demonstrate and confirm the predicted order in these grasses, only seven to ten reconstructed nuclei were required. Presumably this test is suitable for general application to other materials whose simple haploid genomes contain between six and about ten biarmed chromosomes.