Dear SLiM community,
I am currently trying to run a large simulation
that simply takes too long to complete.
I would like to hear if anybody has some suggestions on how to speed
this up.
The aim of my simulation is the following.
Simulate the known demographic history (which includes a severe bottleneck) of
my study system and (a) calculate genetic load of the entire population at
given time points, and (b) perform a viability analysis (i.e.: check how often
my population would go extinct during or after the bottleneck).
Main features of the simulations are:
- Initial population size of 300,000 samples;
- I only simulate deleterious mutations (neutral mutations do not contribute to genetic load), which appear on my ~2Gb genome at a rate of U = 1.2 deleterious mutations per diploid genome per generation (close to empirical estimates for some model species);
- It mimics the reproductive biology of the
species, which causes Ne to be approximately 0.01 of the census size due to
highly skewed reproductive success.
Now, a few considerations:
- Obviously the simulation takes super long because I am simulating a large population. The size of 300,000 was chosen because I do have a Ne estimate of 3,000 (3,000 * 100 = 300,000) and I therefore believe I have to stick to it.
- One way around may be to not simulate the reproductive biology of this species (which causes N/Ne to be 100) and simply input Ne sizes (rather than census sizes) into an ideal WF model. However, this is not desirable because the specific reproductive biology of this species may introduce important stochasticity (not captured by a WF model) which may lead to extinction during the bottleneck. So, I’m not really keen to remove this aspect;
- Scaling down the genome does not appear to improve the situation, because U needs to be 1.2 also within a smaller genome (otherwise the genetic load would be underestimated). Thus, the number of mutations SLiM has to keep track of is the same. Moreover, if I scale down the genome I will have to increase the recombination rate so that the “genetic” length of the genome will be the same (otherwise, linked selection (or, on the other hand “linked drift”) during purging, likely to happen during the bottleneck, will be unrealistically influential).
The main problem is that the simulation requires a
really long burn-in for reaching equilibrium, but the burn-in is clearly
non-neutral. As a consequence, I think cannot use coalescent simulations or
recapitation to generate the burn-in.
I then guess my question is: can anyone suggest a
way to generate a non-neutral burn-in? Importantly, the suggested method should
still allow me to obtain information on selection coefficient, dominance and
allele frequency (relative to the entire population, and not a sample, ideally)
of ALL deleterious mutations present in the population at given time points.
Thank you all very much in advance for your insights and help,
Best,
David
--
SLiM forward genetic simulation: http://messerlab.org/slim/
---
You received this message because you are subscribed to the Google Groups "slim-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to slim-discuss...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/slim-discuss/61ef20bb-8253-4c07-82a3-1a448ee6f44fn%40googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/slim-discuss/CAHJY8dWbPiSrCa7F20t9ra5N1q%3D_NXwxYwaFXEBp9RoWRo5sOw%40mail.gmail.com.