Stochastic Models for Replication Origin Spacings in Eukaryotic DNA Replication

Abstract: We consider eukaryotic DNA replication and in particular the role of replication origins in this process. We focus on origins which are `active' - that is, trigger themselves in the process before being read by the replication forks of other origins. We initially consider the spacings of these active replication origins in comparison to certain probability distributions of spacings taken from random matrix theory. We see how the spacings between neighbouring eigenvalues from certain collections of random matrices has some potential for modelling the spacing between active origins. This suitability can be further augmented with the use of uniform thinning which acts as a continuous deformation between correlated eigenvalue spacings and exponential (Poissonian) spacings. We model the process as a modified 2D Poisson process with an added exclusion rule to identify active points based on their position on the chromosome and trigger time relative to other origins. We see how this can be reduced to a stochastic geometry problem and show analytically that two active origins are unlikely to be close together, regardless of how many non-active points are between them. In particular, we see how these active origins repel linearly. We then see how data from various DNA datasets match with simulations from our model. We see that whilst there is variety in the DNA data, comparing the data with the model provides insight into the replication origin distribution of various organisms.