FFPopSim: an efficient forward simulation package for the evolution of large populations

Area of Science:

• Population genetics
• Computational biology
• Evolutionary dynamics

Background:

• Simulating evolutionary dynamics in large populations with many genetic loci is computationally intensive.
• Standard algorithms face limitations in speed and scalability with increasing numbers of loci (L) or population size (N).

Purpose of the Study:

• To develop novel, efficient algorithms for simulating multi-locus population genetics.
• To overcome the computational bottlenecks of existing simulation methods.

Main Methods:

• Developed algorithms with a run time scaling as 3(L) for simulating sexual populations with multiple crossovers.
• Implemented an analogue of the Fast Fourier Transform (FFT) for efficient computation.
• Included a streamlined individual-based framework with C++ classes and a Python interface.

Main Results:

• Achieved a significant reduction in simulation run time, scaling as 3(L) compared to standard 8(L) algorithms.
• Further reduced run time to L2(L) when restricting the number of crossovers to at most one.
• The new algorithms accommodate arbitrary genetic maps and fitness functions, including epistasis.

Conclusions:

• The developed algorithms provide a computationally efficient tool for multi-locus population genetics.
• Enables more extensive simulations of evolutionary dynamics, particularly for large populations and complex genetic architectures.
• Facilitates research in areas such as quantitative genetics and evolutionary systems biology.