Parameter estimation and SFS generation

[Bushra Shahid]

Dear Laurent and Fsc team,

I have been getting familiar with running fsc28 but have a couple questipns I was hoping you could help me with:

1. One of the models I wish to test is a hybrid model in a 3D SFS where POP1 is a hybrid of POP2 and POP0. This is how my .tpl file looks:

3 historical event
T1 2 1 MIG 1 0 0
T1 1 0 1 RES1 0 0

T2 0 2 1 RES2 0 0

And I set the prior on MIG in the .est file as:

0 MIG unif 1e-10 0.9 output bounded

but I always notice that the MIG rate is always estimated as close to whatever the upper limit is (in this case 0.9). Would you happen to know why this happens? 

2. I also want to generate a multiSFS of 5 pops (20 indv each). I have been using easySFS to generate the sfs. The pops have a mean coverage of 8-15x. Do you know how sensitive fastsimcoal estimations are when you have an SFS generated from pops that vary in their coverage? Alternatively, would you suggest to use genotype likelihoods instead for all pops via ANGSD for SFS generation if I have coverage as low as 8x for 1 pop or is 8x fine coverage but to downproject all pops to 8x?

Thank you for your help!

Best,

Bushra