Under certain conditions, the expectation of VG is 4muVS. These sims check that the code is giving that.
To run:
qsub hpc/validate_additive.sh
#Wait a while
python python process_validate_additive_output.py -i popstats.h5 -o popstats.csv
R --no-save --args popstats.csv popstats.pdf < Rscripts/plot_validate_H2_output.R- How does
$\sigma_\mu$ affect things? - I should probably do a "track everything" simulation, so that I can track pop-gen stats for interesting replicates. Alternately, I should require that the tracking API use a diferernt RNG, which is probably way more efficient. How, though, w/o breaking things?
- How does linkage affect things?
- TFL2013/GBR model
The latter should be done via fwdpp's multi-locus API:
- The total mutation rate is
$\mu$ - There are
$k$ regions, each with mutation rate such that$\sum_i \mu_i = \mu$ . - Each has a within-region recombination rate
- Each has a within-region neutral mutation rate
- The within-region rates should be set to model a 100kb "human" region, b/c that'll give a lot of neutral SNPs to sample
- Vary linkage amongst regions -- 0.1, 0.25, 0.5
- Track everything--samples and mutation frequncies.