Simulated genotypes

We draw \(p = 1000\) allele frequencies:

p = 1000
af = rtrunc(p, spec="beta", .05, .95, shape1=.05, shape2=.05)

• What’s the histogram

An individual can be thought of as a list (data frame or tibble) of three vectors

ind = tibble(pat=rbinom(p, size=1, prob=af),
             mat=rbinom(p, size=1, prob=af),
             dos=pat + mat)

where dos is the dosage of the ALT allele and can take the value 0, 1 or 2.

• Create two individuals;

  • What’s the correlation between their dosage vectors?
  • At how many SNPs do their dosage vectors are in agreement?
  • To derive a statistically meaningful picture, replicate and plot a histogram

• Create a family of three: two parents and one child, assuming that SNPs are independent;

  • What’s the correlation between their dosage vectors?
  • At how many SNPs do their dosage vectors are in agreement?
  • To derive a statistically meaningful picture, replicate and plot a histogram

• Create a family of three: two parents and one child, assuming that SNPs are in full LD;

  • What’s the correlation between their dosage vectors?
  • At how many SNPs do their dosage vectors are in agreement?
  • To derive a statistically meaningful picture, replicate and plot a histogram

• Discuss the two models and their implications and their most obvious limitations

Relatedness in plink

• Create a PED/MAP cohort containing families of three: two parents and a child. As an additional constraint: make sure that you also have cousins in your dataset. Assume independent SNPs and allele frequencies provided by af.

• Analyse the data with plink

• Instructions to install plink on JupyterHub