hello everyone
I'm writing scenarios with fsc26
however, I have trouble filling the sample sizes in tpl files
so far I just fill the sample sizes with *2 of my empirical sampling sizes (cause my focal species is diploid so pop0=181*2, pop1=24*2)
however, these sample sizes tremendously increase the computation time to almost 1 month (I'm using Linux with 10 cores for each model)
to save computation time, will the estimated parameters be strongly affected by the smaller sample sizes I give in my scenario (e.g., only use half the original sample sizes)?
thanks a lot!
below is one of my tpl and est files of an early migration model
//Parameters for the coalescence simulation program: fsimcoal2.exe
2 samples to simulate :
//Population effective sizes (number of genes)
NEW
NHC
//Samples sizes and samples age
362
48
//Growth rates: negative growth implies population expansion
0
0
//Number of migration matrices : 0 implies no migration between demes
3
//Migration matrix 0
0 0
0 0
//Migration matrix 1
0 MIG1
MIG2 0
//Migration matrix 2
0 0
0 0
//historical event: time, source, sink, migrants, new deme size, new growth rate, migration matrix index
2 historical event
TNOMIG 0 0 0 1 0 1
TDIV 1 0 1 RESIZE0 0 2
//Number of independent loci [chromosome]
1 0
//Per chromosome: Number of contiguous linkage Block: a block is a set of contiguous loci
1
//per Block:data type, number of loci, per generation recombination and mutation rates and optional parameters
FREQ 1 0 1.8e-9 OUTEXP
// Search ranges and rules file
// ****************************
[PARAMETERS]
//#isInt? #name #dist.#min #max
//all Ns are in number of haploid individuals
1 NEW logunif 10 1e7 output
1 NHC logunif 10 1e7 output
1 NANC logunif 10 1e7 output
0 MIG1 logunif 0.00001 1 output
0 MIG2 logunif 0.00001 1 output
1 TNOMIG logunif 10 1e5 output
1 TPLUS logunif 1 1e5 output
[RULES]
[COMPLEX PARAMETERS]
0 RESIZE0 = NANC/NEW hide
1 TDIV = TNOMIG+TPLUS output
by the way, this model will stuck sometime and I have to re-simulate the model, could I edit something in this scenario to avoid stuck? thanks for any suggestion!
Cause no one is answering, I will share a paper someone shared with me recently Article Robust Demographic Inference from Genomic and SNP Data
The samples sizes In this paper is 20-40 among models, however, it seems to work quite well for estimating parameters
So maybe it is acceptable to use small samples sizes with fastsimcoal2
- Badges
- Science topic
- Similar topics
- Filing