slow simulation when recombination rate is high

[Devashish13]

Hi,

I am simulating whole genome data for human population. the parameters i used are as follows

number of diploids = 200
mutation rate = 1.25e-8
recombination rate = 8e-7 (estimated using smc++)
sequence length = chromosome length
demographic model = neutral one population model, specified using effecting population size changes obtained through smc++.

for chromosome 22,
recombination rate = 1e-8, time taken = 12 seconds
recombination rate = 1e-7, time taken = 12 minutes
recombination rate = 8e-7 still going on it's been 33 hours as of raising this issue.

High recombination rate will slow the simulation but i have no idea how much time it gonna take.

kindly let me know what is going on here!!
is there any way to make it faster?

thanks,
Devashish

[grahamgower]

Hi @Devashish13,
Yes, this is a known issue (#1651). See Baumdicker et al. 2022 (the msprime-1.0 paper), in particular the "Recombination" section and the "Time complexity of Hudson’s algorithm" section of the Appendix. The latter provides a graph that you should be able to use to estimate the running time.