Simulation and quantitative analysis of spatial centromere distribution patterns

A prominent feature of eukaryotic chromosomes are centromeres, which are specialized regions
of repetitive DNA required for faithful chromosome segregation during cell division. In interphase
cells centromeres are non-randomly positioned in the three-dimensional space of the nucleus in a
cell-type specific manner. The functional relevance and the cellular mechanisms underlying this
observation are unknown, and quantitative methods to measure distribution patterns of
centromeres in 3D space are needed. Here we have developed an analytical framework that
combines robust clustering metrics and advanced modeling techniques for the quantitative analysis
of centromere distributions at the single cell level. To identify a robust quantitative measure for
centromere clustering, we benchmarked six metrics for their ability to sensitively detect changes
in centromere distribution patterns from high-throughput imaging data of human cells, both under
normal conditions and upon experimental perturbation of centromere distribution. We find that
Ripley’s K Score has the highest accuracy with minimal sensitivity to variations in centromeres
number, making it the most suitable metric for measuring centromere distributions. As a
complementary approach, we also developed and validated spatial models to replicate centromere
distribution patterns, and we show that a radially shifted Gaussian distribution best represents the
centromere patterns seen in human cells. Our approach creates tools for the quantitative
characterization of spatial centromere distributions with applications in both targeted studies of
centromere organization as well as in unbiased screening approaches.