Post date: Aug 05, 2017 8:45:40 PM
1. We are including the genomic change/Fst results. All of the code for this is in the popgen* directories within /uufs/chpc.utah.edu/common/home/u6000989/projects/timema_fluct/genomic_change_dark_morph/. For each data set the actual code is in genomicChangeMelStripy.R. We first define a single mel-stripey locus (4139489 on 702.1 to 6414835 on 128). Fst for that regions is calculated as mean(pi_avg-pi_sub)/mean(pi_avg). We then consider only scaffolds with at least as many SNPs as mel-stripey and randomly subset a contiguous set of SNPs of the same numbers as that in mel-stripey; Fst is then calculated the same way. In each case, mel-stripey had the highest p-vaule, thus p = 1/(#scaf+1). For the Ecology Letters experiment is survival for all, Fst is between For FHA Fst, is between 2011 and 2013. For OGA, Fst is between A and C re-samples.
Here are the key statistics
Ecol. Letters: mel-stripe Fst = 0.0030, no. snps 74,305, no. scaffolds = 16, p = 0.059
FHA: mel-stripe Fst = 0.0051, no. snps 780, no. scaffolds = 40, p = 0.024
OGA: mel-stripe Fst = 0.054, no. snps 1180, no. scaffolds = 39, p = 0.025
Fisher's combined probability test: X2 = 20.4976, df = 6, p = 0.0023
2. We are using the relative fitness abc models for Ecology Letters and FHA. The key code is in fhaEcolS.R and fhaAbc6g.R in /uufs/chpc.utah.edu/common/home/u6000989/projects/timema_fluct/genomic_change_dark_morph/sel_fha/ (both require loading R workspaces, fhaAbc6g.rdat for FHA, and this one is ready to go, the other requires most code to be re-ran). I used that R code to calculate the posterior probability that the most fit genotype (from point estimates) was more fit than each other genotype. Here are the numbers:
Ecol. Letters, posterior prob that fitness of s/s > m/u = 0.9782, u/u = 0.8236, m/m = 0.815, m/s = 0.795, u/s = 0.893.
FHA, posterior prob that fitness of s/m > m/u = 0.9038, s/s = 0.9686, u/u = 0.8114, m/m = 0.9224, u/s = 0.474.