Large scale simulation of bacterial population evolution over host contact networks

Abstract

In this work we address the simulation of genetic evolution of bacterial populations in presence of host contact networks. In particular we consider traditional evolution models combined with well mixed and not well mixed host populations, the latter being more realistic. To our knowledge this is the first approach to consider not well mixed host populations, described through complex contact networks. Our results point out that bacterial population diversity can be severely affected by host contact network topology and transmission probabilities alone, without selection phenomena taking place.

Full text

Large scale simulation of bacterial population evolution over host contact
networks
Andreia Sofia Teixeira∗1,2Pedro T Monteiro1,2Alexandre P Francisco1,2Jo˜ao A Carri¸co3
1INESC-ID Lisboa, 2CSE Dept, IST, Universidade de Lisboa, 3UMMI, Instituto de Medicina Molecular, Faculdade de Medicina, Universidade de Lisboa
∗To whom correspondence should be addressed: steixeira@inesc-id.pt
1Abstract
In this work we address the simulation of genetic evolution of bacterial
populations in presence of host contact networks. In particular we con-
sider traditional evolution models combined with well mixed and not well
mixed host populations, the latter being more realistic. To our knowledge
this is the first approach to consider not well mixed host populations, de-
scribed through complex contact networks. Our results point out that
bacterial population diversity can be severely affected by host contact
network topology and transmission probabilities alone, without selection
phenomena taking place.
2Problem
Can strain persistence and local evolutionary events lead to drastic re-
ductions of SID values that can erroneously be attributed to selection
events? What is the impact of host contact network topologies on bac-
terial population diversity, even in the absence of selection phenomena?
3Models
4Network Simulator Parameters
Network Parameters
•Network Structure (Topology)
•Network Dynamics (Individual Exchange Ratio)
•Node Capacity (Number of Individuals per Node)
Evolution Parameters
•Number of locus in allelic profile
•Evolutionary Model (WF or Moran)
•Mutation and Recombination Rates
•Number of Generations
•Exchange Frequency: generations/exchange between nodes
5Results
To the first problem we started to run a simulation using the Karate
network as an example of a weighted graph. Each node had a capacity
for 1000 individuals and the simulation was run for 7 loci, similar to a
typical MLST profile. The mutation/recombination ratio was set at 3/10
and, at each generation of a Wright Fisher Model, 5% the strains were
exchanged through each edge on the graph. The simulation was run for
2500 generation and Simpson’s Index of Diversity (SID) was calculated
for each node. As we can observe there are events of drastic reductions
of SID values without any mechanism of selection. This reflects that
distinction between drift and selection is essential if we aim to understand
natural evolutionary processes.
To the second goal we run three different simulations, with three different
topologies - Clique, Star, Grid. Each network has 400 nodes, undirected
edges and each node has 2500 individuals. The simulations run for 50 000
generations with an exchange frequency of 4 evolutions for 1 exchange.
Observing the three plots we can see that each network has a different
impact in the diversity growth of the bacterial population, reinforcing
what we had mentioned before about the drops of the diversity values
without selection events.
6Final remarks
•Wright-Fisher model based simulator for population evolution on top
of host contact networks.
•Experimental results point out that strain persistence and local evo-
lutionary events reflect local expansion and limitation of genetic ex-
change due to the host network.
•Network topology can then lead to drastic reductions of SID values
that can erroneously be attributed to selection.
7Acknowledgments
This work was partly supported by national funds through
FCT – Funda¸c˜ao para a Ciˆencia e Tecnologia, under projects
UID/CEC/50021/2013.