Parametric Linkage Analysis.

paramlink2

The goal of paramlink2 is to perform parametric linkage analysis in medical pedigrees. Both singlepoint and multipoint LOD scores are supported (the latter requires MERLIN). The package includes tools for visualising LOD scores and summarising peaks for use in downstream analysis.

paramlink2 is part of the ped suite collection of packages for pedigree analysis in R, featured in the book Pedigree Analysis in R (Vigeland, 2021). Chapter 9 of this book gives an introduction to linkage analysis, and includes a detailed worked example using paramlink2.

As hinted to by the package name, paramlink2 is a reinstalment of paramlink, which is no longer actively developed. The two version are not compatible, and all new users are recommended to use paramlink2.

Installation

To get the current official version of paramlink2, install from CRAN as follows:

install.packages("paramlink2")

Alternatively, you can obtain the latest development version from GitHub:

# install.packages("devtools") # install devtools if needed
devtools::install_github("magnusdv/paramlink2")

An example

library(paramlink2)
#> Loading required package: pedtools

The family below is affected with an autosomal dominant disorder. The built-in dataset dominant1 contains genotypes for 14 of the members, at 248 SNP markers on chromosome 1. We will perform parametric linkage analysis on the dataset, hoping to identify a genomic region linked to the disease locus.

The dataset contains three elements: ped, aff and map. For simplicity we store these in separate variables:

ped = dominant1$ped
aff = dominant1$aff
map = dominant1$map

The pedigree plot above was produced with the command

plot(ped, aff = aff, starred = typedMembers)

In order to compute LOD scores, we need to provide a disease model. This is conveniently done with the function diseaseModel(). For this example we use a fully penetrant autosomal dominant model:

modAD = diseaseModel("AD")

modAD
#> Autosomal inheritance
#> Penetrance: (f0, f1, f2) = (0, 1, 1)
#> Disease allele frequency: 1e-05

The diseaseModel() function contains a number optional arguments which can be used to create more realistic models, including phenocopies, reduced penetrance and liability classes. Consult the documentation ?diseaseModel for details about these!

Singlepoint LOD scores for the family are now computed as follows:

lods = lod(ped, aff = aff, model = modAD)

The summary() function prints the highest score:

summary(lods)
#> Max LOD score: 2.533179 
#> Achieved at marker(s): m47

Furthermore, plot() produces a nice graph of the LOD scores:

plot(lods)

As typical for singlepoint scores, the graph is quite noisy and not easy to interpret, although we see indications of a peak. A cleaner picture may be obtained by multipoint analysis, which is available if you have MERLIN installed on your computer. A special wrapper, merlinLod() takes care of all the input and output files to MERLIN, so that we never have to leave R. Moreover, the syntax is similar to that of lod() except that we may add a linkage map of the markers.

lodsM = merlinLod(ped, aff = aff, model = modAD, map = map)

The following command plots the MERLIN scores together with the singlepoint scores:

plot(lods, col = 8)
points(lodsM, col = 2)
legend("topright", c("Multi", "Single"), col = c(2, 8), lwd = 2)

This graph shows a convincing peak of LOD = 3, which is close to the traditional significance threshold LOD = 3.3 for AD disorders. We print some details about the peak and its location:

peakSummary(lodsM, threshold = 2)
#>   CHROM FROM  TO  N FROM_MB TO_MB LEN TELO MAXLOD
#> 1     1  m34 m56 23      34    56  22        3.01

In other words, the disease locus is most likely somewhere on chromosome 1 between 34 Mb and 56 Mb.