Post date: Jun 15, 2015 2:24:41 AM
#Based on the change in allele freq (amount of evolutionary change) from generation to generation (assuming all attributable to drift; neutral model); btwn two generations, similar to FST
#Different from basing change in het.: looking at allele freq change for each locus, taking all info to get Ne
#Need one infile per group: 2 columns, 1st allele freq, 2nd 2N (separated by one space)
#Used bayesian pop model allele frequencies, 13999 loci
#2Ns: C-EL: 100, C-F1: 126, C-F2: 124, C-F3: 22
#sampling scheme: we sampled and put them back; correction needs census size (did we sample them all?):
Number of captives on Feb 22, 2013:
founders: 57
F1s: 381
F2s: 310
F3s: 11
------
## founder to F3 ***maybe the best estimate because more confident with more generations; assumes constant Ne between these two time points; reflects gravid ones (but doesn't reflect current / future pop because there won't continue to be outside contributions)
../varne -a Ne_p_popaf_E-BS-C-EL.txt -b Ne_p_popaf_E-BS-C-F3.txt -l 13999 -t 3 -n 57 -x 1000
Reading input from files: Ne_p_popaf_E-BS-C-EL.txt and Ne_p_popaf_E-BS-C-F3.txt
Estimating effective population size
hat{Ne} = 217.945
median of posterior = 218.103
50% credible intervals = 197.847, 244.44
90% credible intervals = 174.613, 290.046
95% credible intervals = 167.155, 303.036
## founder to F1 **
../varne -a Ne_p_popaf_E-BS-C-EL.txt -b Ne_p_popaf_E-BS-C-F1.txt -l 13999 -t 1 -n 57 -x 1000
Reading input from files: Ne_p_popaf_E-BS-C-EL.txt and Ne_p_popaf_E-BS-C-F1.txt
Estimating effective population size
hat{Ne} = 171.121
median of posterior = 171.177
50% credible intervals = 160.921, 181.569
90% credible intervals = 148.537, 200.777
95% credible intervals = 144.59, 206.781
## F1 to F2
../varne -a Ne_p_popaf_E-BS-C-F1.txt -b Ne_p_popaf_E-BS-C-F2.txt -l 13999 -t 1 -n 381 -x 1000
Reading input from files: Ne_p_popaf_E-BS-C-F1.txt and Ne_p_popaf_E-BS-C-F2.txt
Estimating effective population size
hat{Ne} = 292.724
median of posterior = 294.344
50% credible intervals = 270.596, 325.357
90% credible intervals = 241.529, 384.924
95% credible intervals = 232.053, 406.785
## F2 to F3
../varne -a Ne_p_popaf_E-BS-C-F2.txt -b Ne_p_popaf_E-BS-C-F3.txt -l 13999 -t 1 -n 310 -x 1000
Reading input from files: Ne_p_popaf_E-BS-C-F2.txt and Ne_p_popaf_E-BS-C-F3.txt
Estimating effective population size
hat{Ne} = 121.736
median of posterior = 122.002
50% credible intervals = 105.719, 145.363
90% credible intervals = 86.6703, 194.243
95% credible intervals = 82.2628, 215.165
## F1 to F3 ***maybe the best estimate because more confident with more generations; assumes constant Ne between these two time points (best idea of where the population is going...)
../varne -a Ne_p_popaf_E-BS-C-F1.txt -b Ne_p_popaf_E-BS-C-F3.txt -l 13999 -t 2 -n 381 -x 1000
Reading input from files: Ne_p_popaf_E-BS-C-F1.txt and Ne_p_popaf_E-BS-C-F3.txt
Estimating effective population size
hat{Ne} = 99.341
median of posterior = 99.3836
50% credible intervals = 92.7461, 106.594
90% credible intervals = 85.0422, 117.83
95% credible intervals = 82.5583, 122.109
----
#16june15 -- it turns out when you take a subset of the first half of the loci when running the model vs the last half, you decrease and increase the estimates, respectively. my allele frequencies are structured in the file and we don't know why -- maybe the way seqman calls things. so, we took the 1,543 high coverage loci (see hicovindexes.txt) we've used in other analyses... nevermind, too few loci. it doesn't really matter that their ordered.
#then took a random subset (8000 loci) increased CIs but didn't change the point estimates much. see euryceaNE.txt for these results. so, use the original results.