In eriqande/SNPcontam: Detecting contaminated samples from SNP genotypes

1. Prepare Data

The MCMC simulation function requires that the input data be a matrix with columns equal to the number of alleles, which is 2N where N is the number of loci, and with rows equal to the number of individuals. The MCMC simulation uses functions from the package fullsniplings to turn a data frame of SNP alleles inot values of 0, 1, 2, or NA. Below is subset of the data frame in its original form:

library(SNPcontam)
load("~/Hollings/SNPcontam/data/swfsc_chinook_baseline.rda")
tmp_data <- swfsc_chinook_baseline
tmp_data <- tmp_data[tmp_data$Pop == "Feather_H_sp", ]
tmp_data[(1:6),(1:6)]

Any columns or rows contain non-SNP information (i.e. information regarding the origin of the individual) should only be removed before using the function MCMC_sims. This is shown below:

data <- tmp_data[,-(1:4)]
data[(1:4),(1:4)]

2. Run the MCMC Simulation Function

Once the data is prepared, the function MCMC_sims can be used to run n repitions of the the function contam_MCMC over a set of simulated data. Below the simulation is run twice with four different values for number of loci (Lvals) and five different values for proportion of contaminated samples (rhovals). The object similuation will be a list containing data frames with the posterior means of the allele frequencies, z values, and contamination proportions that are used to create plots using MCMC_allplots.

library(fullsniplings)
library(parallel)
simulation <- MCMC_sims(sample_data = data,N = 200,Lvals = c(20,60,100,200),rhovals = c(0,.025,0.075,0.2,0.5), n = 2 )

3. Make Plots

As stated above the function MCMC_allplots is used to make three plots and one table. The code is shown below. The rho and loci values correspond with the graph of the allele frequency estimates and dictate the subset of data used to make the graph.

```r library(ggplot2) library(xtable) library(grid) library(gridExtra) MCMC_allplots(sim = simulation,rho = .2,loci = 20)

eriqande/SNPcontam documentation built on May 16, 2019, 8:44 a.m.