Math Biology Seminar: Caroline Colijn

   
Ecological competition between strains of a pathogen occurs when strains compete for hosts -- either for susceptible hosts, host resources during co-infection, or the ability to re-infect hosts. Competition is important because when strains compete with each other, intervening against only some of them can pave the way for rises in others. This has happened, for example, with the introduction of polyvalent vaccines against Streptococcus pneumoniae. However, detecting ecological competition between strains of an infection is very challenging, because competition is by its nature revealed over relatively long periods of time and is a population-level phenomenon which we would not expect to observe in small-scale studies. Even population-level dynamical (ODE) models, which are frequently used in such situations, are hard to formulate and calibrate. Indeed, such models often make hidden assumptions about competition, rather than aiding in its estimation. I have therefore been motivated to ask: can sequence data for pathogens allow us to detect ecological competition? Large and rich datasets of pathogen gene sequences are now available, due to the development of next-generation sequencing; perhaps they can be of assistance if appropriately linked to models with and without competition. Here, I present a dynamical model in which there is a competition parameter which ranges continuously from 0 (where pathogen strains are independent of each other) to 1 (where competition is complete, and strain dynamics show competitive exclusion). It predicts that the branching rates in phylogenies for competing strains should be anti-correlated. A stochastic implementation of the model gives rise to pathogen phylogenies that are quantitatively different, both in their structures and their branch lengths, from phylogenies without competition. This leads to a distinct profile for a phylogeny under ecological competition: such trees have high imbalance early in the tree, greater topological distances from the root to the tips, lower widths and a characteristic skew in inter-branch distances, among other properties. I analyse a phylogeny of within-host HIV sequences and show that it fits the profile of ecological competition. I conclude with a discussion of other organisms and future directions for this work.