SLiM 3: Forward genetic simulations beyond the Wright–Fisher model

Abstract

With the desire to model population genetic processes under increasingly realistic scenarios, forward genetic simulations have become a critical part of the toolbox of modern evolutionary biology. The SLiM forward genetic simulation framework is one of the most powerful and widely used tools in this area. However, its foundation in the Wright-Fisher model has been found to pose an obstacle to implementing many types of models; it is difficult to adapt the Wright-Fisher model, with its many assumptions, to modeling ecologically realistic scenarios such as explicit space, overlapping generations, individual variation in reproduction, densitydependent population regulation, individual variation in dispersal or migration, local extinction and recolonization, mating between subpopulations, age structure, fitness-based survival and hard selection, emergent sex ratios, and so forth. In response to this need, we here introduce SLiM 3, which contains two key advancements aimed at abolishing these limitations. First, the new non-Wright-Fisher or "nonWF" model type provides a much more flexible foundation that allows the easy implementation of all of the above scenarios and many more. Second, SLiM 3 adds support for continuous space, including spatial interactions and spatial maps of environmental variables. We provide a conceptual overview of these new features, and present several example models to illustrate their use. These two key features allow SLiM 3 models to go beyond the Wright-Fisher model, opening up new horizons for forward genetic modeling.

Subsets of NLR genes show differential signatures of adaptation during colonization of new habitats

“…Using forward simulations (Notes S5), we tested that genes under different selective pressures in different populations can be revealed by outlier high F ST values compared with the neutral expected distributions from our neutral demographic scenario. For that, we ran simulations using SLIM (Haller & Messer, 2019), assuming genes evolving neutrally in all populations and changing from neutral to either positive or balancing selection in the southward colonization processes.…”

Tree‐sequence recording in SLiM opens new horizons for forward‐time simulation of whole genomes

“…Since computational resources are finite, this can often make it difficult or, in practical terms, impossible to run some models. Advances in computing power have gradually extended the boundaries of what is possible, as have performance improvements due to improved forward simulation software (Haller & Messer, , ; Messer, ; Thornton, ), but computational overhead continues to hold back progress in the field by limiting the level of realism that can be attained in models.…”

Are populations like a circuit? Comparing isolation by resistance to a new coalescent‐based method

“…Development of new methods for continuous geography and assessment of how current methods behave when faced with continuous geography is an important challenge facing the field. This has been made substantially easier with the introduction of continuous space into the simulator, SLiM (Haller & Messer, ), that we used to test the behavior of our method with more realistic data. Another promising approach centres around the Spatial Lambda‐Fleming‐Viot model (Barton, Etheridge, & Véber, ), which provides a mechanistic model in which coalescent simulations are possible.…”

“…To produce realistic data, we implemented forwards‐time simulations using SLiM v3.1 (Haller, Galloway, Kelleher, Messer, & Ralph, ; Haller & Messer, ), with individuals living across continuous, two‐dimensional geography (sometimes with barriers), from which we recorded genomic data. The basic simulation, which we modified to produce several other situations, is as follows.…”