In richelbilderbeek/plinkr: Interface to PLINK

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

Introduction

In this vignette, we are going to detect associations between genetic information and one or more quantitative traits.

We will be using simulated data, with a simple relationship between genotype and phenotype. To see how this same analysis is done with more messy data, see the 'test_assoc_qt' vignette.

Setup

Set the RNG seed:

set.seed(11)

First load plinkr:

library(plinkr)

This vignette will build whether or not PLINK is installed

if (is_plink_installed()) {
  message("PLINK is installed")
} else {
  message(
    "PLINK is not installed",
    "",
    "Tip: run 'plinkr::install_plinks()' to do so"
  )
}

Detecting associations for quantitative traits

To do an association (both for a single and multiple traits), we need some data:

• The .map table or mapping table, which contains the location of the single-nucleotide polymorphisms (SNPs)
• The .ped table or pedigree table, which contains the pedigree of the individuals and their SNP values
• The phenotype table, which contains the one or more phenotypes of the individuals

Here, we create some simple data, as to be used in testing, or -in this case- a simple demonstration:

assoc_qt_data <- create_demo_assoc_qt_data(
  n_individuals = 16
)

The mapping table

knitr::kable(assoc_qt_data$data$map_table)

The pedigree table

The PLINK example \code{.ped} contains:

• the pedigree of the individuals
• the SNP calls of the individuals

Show only the pedigree:

knitr::kable(assoc_qt_data$data$ped_table[, 1:6])

Note that the pedigree table has a column called case_control_code, which will be completely ignored.

Show only the SNP calls:

knitr::kable(assoc_qt_data$data$ped_table[, -(1:6)])

The phenotype table

The pedigree table has a column called case_control_code, which will be completely ignored, as it only allows to put in one phenotype. Instead, we use a table of one or more phenotypic values:

knitr::kable(assoc_qt_data$phenotype_data$phe_table)

The phenotypes are named after their relationship to the genotype:

• ´random´: there is no relationship, values are random
• ´additive´: there is a perfect additive relationship, with a starting value of 10.0, where each 'T' increases the trait by 0.5
• ´epistatic´: there is an epistatic phenotype, with a starting value of 20.0, that can only be 21.0 if all four nucleotides on four SNPs are the uncommon alleles

Detecting the association

With the mapping, pedigree and phenotype table, we can detect the association between genotype and the two traits:

if (is_plink_installed()) {
  assoc_qt_result <- assoc_qt(
    assoc_qt_data = assoc_qt_data
  )
  knitr::kable(assoc_qt_result$qassoc_table)
}

• trait_name: name of the quantitive trait,
• CHR: Chromosome number
• SNP: SNP identifier
• BP: Physical position (base-pair)
• NMISS: Number of non-missing genotypes
• BETA: Regression coefficient
• SE: Standard error
• R2: Regression r-squared
• T: Wald test (based on t-distribution)
• P: Wald test asymptotic p-value

In the second row, the association between snp_2 and the additive phenotype is discovered correctly, as the R2 (the R-squared value, i.e. the proportion of the phenotype that can be explained by the genotype) gives the highest possible value of 1.0, which denotes the phenotype can be perfectly explained by the genotype.

Add a bit of noise to the data

assoc_qt_data$phenotype_data$phe_table$additive <-
  assoc_qt_data$phenotype_data$phe_table$additive + 
  runif(n = 16, min = -0.001, max = 0.001
)

if (is_plink_installed()) {
  assoc_qt_result <- assoc_qt(
    assoc_qt_data = assoc_qt_data
  )
  knitr::kable(assoc_qt_result$qassoc_table)
}

Cleanup

clear_plinkr_cache()

check_empty_plinkr_folder()

richelbilderbeek/plinkr documentation built on March 25, 2024, 3:18 p.m.