This directory contains all data, code, and output pertinent to the paper "Bayesian Inference of Selection in the Wright-Fisher Diffusion Model". Details regarding the specific files are provided below.

Data:

Data/fly2Ln.txt

• Text file containing the ending allele frequencies for the entire 2L chromosome for the neutral fly data

Data/fly2Ls.txt

• Text file containing the ending allele frequencies for the entire 2L 2L chromosome for the hypoxic fly data

Data/fly-n.txt

• Text file containing the ending allele frequencies for the subset of the 2L chromosome identified by Ronen et al (2013) for the neutral fly data

Data/fly-s.txt

• Text file containing the ending allele frequencies for the subset of the 2L chromosome identified by Ronen et al (2013) for the hypoxic fly data

Data/nlocs2L.txt

• Text file containing SNP locations for the entire 2L chromosome for the neutral fly data

Data/slocs2L.txt

• Text file containing SNP locations for the entire 2L chromosome for the hypoxic fly data

Code:

WrightFisherSampling.cu

• CUDA code for simulating the Wright-Fisher diffusion using the GPU
• Produces "qSamples/final_samples_cuda.txt"

WFwithSelection-ApproxAll.py

• Python code for simulating a general Wright-Fisher diffusion
• Uses a Normal approximation to compute m within the exact algorithm of Jenkins and Spano (2015)
• Calls "NeutralWF-NormApproxAll.py"

NeutralWF-NormApproxAll.py

• Python code for simulating a neutral Wright-Fisher diffusion
• Uses a Normal approximation to compute m within the exact algorithm of Jenkins and Spano (2015), even when t is not too small
• Calls "NormalApprox.py"

NormalApprox.py

• Python code to compute Normal approximation for m within the exact algorithm of Jenkins and Spano (2015)

ComputeKDEs.py

• Python code for computing kernel density estimates
• Reads in "qSamples/final_samples_cuda.txt"
• Produces "DensityEstimates/kdes_fly1217"

MCMC_simXX.py XX=0,55,11

• Python code for generating simulated data with s=0.0, s=5.5, or s=11.0 then sampling from the posterior distribution via MCMC
• Uses "informative" prior for initial allele frequencies
• Calls "WFwithSelection-ApproxAll.py"
• Reads in "DensityEstimates/kdes_fly1217" and "Data/fly-n.txt"
• Produces "Chains/MCMC_simXX_ssX.txt"

MCMC_simXX_unif.py XX=0,55,11

• Python code for generating simulated data with s=0.0, s=5.5, or s=11.0 then sampling from the posterior distribution via MCMC
• Uses Uniform prior for initial allele frequencies
• Calls "WFwithSelection-ApproxAll.py"
• Reads in "DensityEstimates/kdes_fly1217" and "Data/fly-n.txt"
• Produces "Chains/MCMC_simXX_unif_ssX.txt"

MCMC_fly_neutral.py

• Python code for sampling from the posterior distribution for the neutral fly data using MCMC
• Calls "WFwithSelection-ApproxAll.py"
• Reads in "DensityEstimates/kdes_fly1217" and "Data/fly-n.txt"
• Produces "Chains/MCMC_fly1n_output.txt"

MCMC_fly_select.py

• Python code for sampling from the posterior distribution for the hypoxic fly data using MCMC
• Calls "WFwithSelection-ApproxAll.py"
• Reads in "DensityEstimates/kdes_fly1217", "Data/fly-n.txt", and "Data/fly-s.txt"
• Produces "Chains/MCMC_fly1s_output.txt"

MCMC_fly_windows.py

• Python code for sampling from the posterior distribution for sliding windows along the 2L chromosome of the hypoxic fly data using MCMC
• Calls "WFwithSelection-ApproxAll.py"
• Reads in "DensityEstimates/kdes_fly1217", "Data/fly2Ln.txt", "Data/nlocs2L.txt", "Data/fly2Ls.txt", and "Data/slocs2L.txt"
• Produces "Chains/MCMC_fly2L_windows.txt"

figure1.R

• R code for generating histograms of the posterior of s for the simulations that use an informative prior for the initial allele frequencies
• Reads in "Chains/MCMC_simXX_ssX.txt" for XX=0,55,11 and X=1,2,3,4
• Produces "Figures/figure1.pdf"

figure2.R

• R code for generating histograms for the posterior of s for the simulations that use an informative prior for the initial allele frequencies
• Reads in "Chains/MCMC_simXX_unif_ssX.txt" for XX=0,55,11 and X=1,2,3,4
• Produces "Figures/figure2.pdf"

figure3.R

• R code for generating histograms for the posterior of s for the fly data
• Reads in "Chains/MCMC_fly1s_output.txt" and "Chains/MCMC_fly1n_output.txt"
• Produces "Figures/figure3.pdf"

figure4.R

• R code for computing and plotting 99% credible intervals for s for sliding windows along the 2L chromosome of the hypoxic fly population
• Reads in "Data/slocs2L.txt" and "Chains/MCMC_fly2L_windows.txt"
• Produces "Figures/figure4.pdf"

Output:

qSamples/final_samples_cuda.txt

• Text file containing simulated values of q from "WrightFisherSampling.cu"

DensityEstimates/kdes_fly1217

• Pickle file containing the kernel density estimates from "ComputeKDEs.py"

Chains/MCMC_simXX_ssX.txt XX=0,55,11 X=1,2,3,4

• Text file containing Markov chains for s for simulated data and an informative prior for the initial allele frequencies

Chains/MCMC_simXX_unif_ssX.txt XX=0,55,11 X=1,2,3,4

• Text file containing Markov chains for s for simulated data and a Uniform prior for the initial allele frequencies

Chains/MCMC_fly1n_output.txt

• Text file containing the Markov chain for (s,q) for the neutral fly data

Chains/MCMC_fly1s_output.txt

• Text file containing the Markov chain for (s,q) for the hypoxic fly data

Chains/MCMC_fly2L_windows.txt

• Text file containing Markov chains for s for sliding windows along the 2L chromosome of the hypoxic fly data

Figures/figureX.pdf X=1,2,3,4

• PDF file containing Figure X from the paper