Almost all sites have multiple mutations when bootstrapping

[Yasuto ISHII]

Hi fastsimcoal development team and users,

I tried to generate parametric bootstrap SFS in order to estimate CI but have some trouble.

My trouble is:

almost all sites have multiple mutations and SFS does not seem infromative.

I followed parametric bootstrap approach described in the manual.

My .par file for estimating SFS is attached (nogf_maxL.par).

And, the command to get SFS from the par file is:

fsc27 -i nogf_maxL.par -n2 -j -d -s0 -x –I -q -B 40 -c 40

The generated SFS files are also attached (*.obs).

nogf_maxL_DAFpop0.obs said 1,489,980 simulated sites had multiple mutations.

I simulated 1,490,000 sites (149 bp x 10,000 independent loci), so it said almost all sites have multiple mutations.

SFS seems to be calculated based on only 20 sites and not informative.

How can I resolve the issue?

Best,

Yasuto

[Laurent Excoffier]

Hi,

i think you have a super large ancestral size of 3826658 x 1030 haploids (about 4 billion haploids or 2 billion diploids), meaning the MRCA will be on average 8 billion generations old. Over this long evolutionary time, mutations will certainly occur at all the 14.9 million bp you simulate, given that you expect several dozens mutations per site given your mutation rate

Check your estimation files and try to understand why you estimate such a large ancestral population size...

best

laurent

[Yasuto ISHII]

HI Laurent,

Thank you for your quick reply.

I attached input files for the point estimation.

The prior distribution for population size is uniform distribution from 1e3 to 1e7.

I just supposed a bifurcation tree and estimated the divergence time of the two lineages.

Could you guess how I can resolve the issue?

Should I change to a narrower distribution, say from 1e3 to 1e6?

Best,

Yasuto

2024年2月26日月曜日 0:56:58 UTC+9 laurent....@gmail.com: