Pedigree Analysis and Familial Aggregation
29 October 2019
## Loading required package: kinship2## Loading required package: Matrix## Loading required package: quadprog## Loading required package: igraph##
## Attaching package: 'igraph'## The following objects are masked from 'package:stats':
##
## decompose, spectrum## The following object is masked from 'package:base':
##
## union##
## Attaching package: 'FamAgg'## The following object is masked from 'package:igraph':
##
## cliques## The following object is masked from 'package:kinship2':
##
## pedigreePackage: FamAgg
Authors: J. Rainer, D. Taliun, C.X. Weichenberger
Modified: 2019-10-29 16:29:36
Compiled: Tue Oct 29 21:41:57 2019
1 Introduction
This package provides basic pedigree analysis and plotting utilities as well as a variety of methods to evaluate familial clustering of cases from a given trait. Identification of families or groups of individuals within families with significant aggregation of cases can aid also in the selection of interesting and promising individuals for whole genome or exome sequencing projects.
For kinship coefficient calculations and pedigree plotting the package relies
and extends the functionality of the kinship2 package [1].
If you use this package please cite Rainer et al. [2].
2 Basic pedigree operations
In the examples below we perform some simple pedigree operations, such as
plotting the pedigree for an individual or family, finding the closest common
ancestor for a set of individuals in a pedigree or retrieving the identifiers
(IDs) of all ancestors for an individual. Basic pedigree information is stored
in FAData objects, thus we first generate such an object from a subset of the
Minnesota Breast Cancer Study provided by the kinship2 package. In the example
below, we generate the FAData providing a data.frame with the pedigree data,
alternatively, the pedigree information could be imported from a file (see
Section 3). Upon data set creation the kinship matrix (i.e. a
matrix containing the kinship coefficient between each pair of individuals in
the whole pedigree) is internally calculated using the functionality from the
kinship2 package [1].
library(FamAgg)
data(minnbreast)
## Subsetting to only few families of the whole data set.
mbsub <- minnbreast[minnbreast$famid %in% 4:14, ]
mbped <- mbsub[, c("famid", "id", "fatherid", "motherid", "sex")]
## Renaming column names.
colnames(mbped) <- c("family", "id", "father", "mother", "sex")
## Defining the optional argument age.
endage <- mbsub$endage
names(endage) <- mbsub$id
## Create the object.
fad <- FAData(pedigree = mbped, age = endage) We can access all the pedigree information stored in this object using the
pedigree method, but also using $. The row names of the pedigree
data.frame as well as the names of the vectors returned by $ are the IDs of
the individuals in the pedigree.
## Use the pedigree method to access the full pedigree
## data.frame,
head(pedigree(fad))## family id father mother sex
## 1 4 1 NA NA M
## 2 4 2 NA NA F
## 3 4 3 25 4 F
## 4 4 4 1 2 F
## 5 4 5 1 2 M
## 6 4 6 1 2 M## or access individual columns using $.
## The ID of the father (0 representing "founders"):
head(fad$father)## 1 2 3 4 5 6
## NA NA 25 1 1 1## Mother:
head(fad$mother)## 1 2 3 4 5 6
## NA NA 4 2 2 2## Sex:
head(fad$sex)## 1 2 3 4 5 6
## M F F F M M
## Levels: M F## We can also access the age of each individual, if
## provided.
head(age(fad)) ## 1 2 3 4 5 6
## NA 78.05886 55.50000 48.00000 75.00342 53.63997To extract the pedigree for a single family we can use the family method,
specifying either the ID of the family or the ID of an individual in the family.
## Extract the pedigree information from family "4"...
nrow(family(fad, family = 4))## [1] 43head(family(fad, family = 4))## family id father mother sex
## 1 4 1 NA NA M
## 2 4 2 NA NA F
## 3 4 3 25 4 F
## 4 4 4 1 2 F
## 5 4 5 1 2 M
## 6 4 6 1 2 M## ...which is the same as extracting the family pedigree
## for an individual of this family.
head(family(fad, id = 3))## family id father mother sex
## 1 4 1 NA NA M
## 2 4 2 NA NA F
## 3 4 3 25 4 F
## 4 4 4 1 2 F
## 5 4 5 1 2 M
## 6 4 6 1 2 M## Note that IDs are internally always converted to character,
## thus, using id=3 and id="3" return the same information.
head(family(fad, id = "3")) ## family id father mother sex
## 1 4 1 NA NA M
## 2 4 2 NA NA F
## 3 4 3 25 4 F
## 4 4 4 1 2 F
## 5 4 5 1 2 M
## 6 4 6 1 2 MAlternatively, we could subset the FAData to individuals of a single family.
## Subset the object to a single family.
fam4 <- fad[fad$family == "4", ]
table(fam4$family) ##
## 4
## 43To explore this family we can plot its pedigree. By default, the plotting
capabilities of the kinship2 package are used to plot pedigrees, but
alternatively, if all required dependencies are available, the HaploPainter
[3] perl script (http://haplopainter.sourceforge.net/) can be
used instead. The switchPlotfun function can be used to switch the plotting
back-end. Available arguments are ks2paint and haplopaint for kinship2 and
HaploPainter plotting, respectively. Note however, that HaploPainter only
allows to export plots to a file, while kinship2 plotting allows, in addition
to export the plot, also to show it as a standard R plot.
Below we use the switchPlotfun to ensure the use of kinship2 plotting
(usually not required) and plot the full available pedigree of individual 3.
If the age of individuals is available, it will be plotted below the
individual’s ID.
switchPlotfun("ks2paint")
## By supplying device="plot", we specify that we wish to visualize the
## pedigree in an R plot. This is the default for "ks2paint", anyway.
plotPed(fad, id = 3, device = "plot") 
The pedigree for an individual or a list of individuals can be extracted using
the buildPed method. By default the method first tries to identify all parents
up to 3 generations in the pedigree, and subsequently all children of the
individuals and all identified parents.
## Build the pedigree for individual 3.
fullPed <- buildPed(fad, id = "3")
nrow(fullPed) ## [1] 29Alternatively, we can extract the smallest possible pedigree for a list of
individuals by specifying prune=TRUE. Internally, the function transforms the
pedigree into a graph, tries to find all paths between the individuals and
returns the sub-graph of all individuals along with individuals along the paths
between them.
## Find the subpedigree for individuals 21, 22 and 17.
buildPed(fad, id = c(21, 22, 17), prune = TRUE) ## family id father mother sex
## 3 4 3 25 4 F
## 4 4 4 1 2 F
## 1 4 1 NA NA M
## 8 4 8 1 2 F
## 17 4 17 28 8 M
## 21 4 21 24 3 M
## 22 4 22 24 3 F
## 2 4 2 NA NA F
## 25 4 25 NA NA M
## 28 4 28 NA NA M
## 24 4 24 NA NA MAnd the pedigree plot for that subset of the whole family:
plotPed(fad, id = c(21, 22, 17), prune = TRUE) 
Note that the pedigree returned by the buildPed method for an individual might
be different than the pedigree of a whole family. The pedigree returned by
buildPed contains only individuals that share kinship with the specified
individual. To exemplify this, we plot the pedigree for the family 14 in the
Minnesota Breast Cancer data set. Note that the individuals in the pedigree plot
depicted as diamonds are individuals with unknown gender. (The message “Did not
plot…” is issued by the kinship2 plotting function and indicates singletons
that are assigned to the family but do neither have parents nor children.)
plotPed(fad, family = "14", cex = 0.4) 
## Did not plot the following people: 457 463 470 471 26067 26068 26098 26099In this family, founder 441 is the founder of two family branches. Building
the pedigree for individual 440 will not include any of the individuals of the
second branch, as he does not share kinship with any of them. The pedigree built
for 447 on the other hand contains also individuals from the second branch as
she shares kinship with them (via her mother 441).
## Check if we have individual 26064 from the second branch in the pedigree
## of individual 440.
any(buildPed(fad, id = "440")$id == "26064")## [1] FALSE## What for the pedigree of 447?
any(buildPed(fad, id = "447")$id == "26064") ## [1] TRUEA family pedigree may consist of many founder couples (i.e. individuals for
which neither father nor mother is defined in the pedigree). To identify the
pedigree’s founder couple (being the couple with the largest number of offspring
generations in the pedigree) the findFounders method can be used. Note that
the function returns always only one couple, even if there might be two founder
couples in the family pedigree with the same number of offspring generations.
## Find founders for family 4.
findFounders(fad, "4") ## [1] "1" "2"Alternatively, it might be of interest to determine the closest common ancestor
between individuals in a pedigree. Below we use the getCommonAncestor method
to identify the common ancestor for individuals 21, 22 and 17 (which we
know from the pedigree a bit above are 1 and 2).
## Find the closest common ancestor.
getCommonAncestor(fad, id = c(21, 22, 17)) ## [1] "1" "2"Other useful methods are getChildren, getAncestors and getSiblings, that
return the children (or all offspring generations up to a specified level), the
parents (or all ancestors) or the siblings for the specified individuals,
respectively.
## Get the children of ID 4.
getChildren(fad, id = "4", max.generations = 1)## [1] "3"## Get the offsprings.
getChildren(fad, id = "4")## [1] "3" "21" "22" "23"## Get all ancestors.
getAncestors(fad, id = "4")## [1] "1" "2"## Get the siblings.
getSiblings(fad, id = c("4")) ## [1] "4" "5" "6" "7" "8" "9" "10"In the whole Minnesota Breast Cancer data set there are 426 families corresponding to 426 founders that had cancer during the screening phase between 1944 and 1952. In the code block below we identify the affected founders per family.
## Add the trait information to the FAData object.
cancer <- mbsub$cancer
names(cancer) <- as.character(mbsub$id)
trait(fad) <- cancer
## Identify the affected founders.
## First all affected individuals.
affIds <- affectedIndividuals(fad)
## Identify founders for each family.
founders <- lapply(unique(fad$family), function(z){
return(findFounders(fad, family = z))
})
names(founders) <- unique(fad$family)
## Track the affected founder.
affFounders <- lapply(founders, function(z){
return(z[z %in% affIds])
})
## Interestingly, not all founders are affected! It seems in some cases
## parents of the affected participants in the screening phase have also
## been included.
affFounders <- affFounders[unlist(lapply(affFounders, length)) > 0]
## The number of families analyzed.
length(founders)## [1] 10## The number of families with affected founder.
length(affFounders)## [1] 2Unexpectedly, only in few families one of the founders is affected. For the other families additional (unaffected) ancestors might have been added at a later time point.
Next we get the number of affected individuals that are related to these affected founders.
kin2affFounders <- shareKinship(fad, unlist(affFounders))
## How many of these are affected?
sum(kin2affFounders %in% affIds)## [1] 7## How many affected are not related to an affected founder?
sum(!(affIds %in% kin2affFounders))## [1] 212.1 Pedigree analysis methods
In this section we perform some more advanced pedigree operations. First, we
identify all individuals in the pedigree that share kinship with individual 4.
## Get all individuals sharing kinship with individual 4.
shareKinship(fad, id = "4") ## [1] "1" "2" "3" "4" "5" "6" "7" "8" "9" "10" "11" "12" "13" "14"
## [15] "15" "16" "17" "18" "19" "20" "21" "22" "23"Next, we determine generations within the pedigree. Generations can only be
estimated for a single family, since in most instances e.g. the year of birth is
not available. Thus, generations are estimated considering the relation between
individuals, starting from the founder couple, i.e. generation 0, assigning
generation 1 to their children and all the mates of their children and so
on. The estimateGenerations method calculates such generation numbers for each
family defined in the object (or for a single family, if the family ID is
provided). The result is returned as a list with the list names corresponding to
the family ID and the list elements being the estimated generation numbers (with
names corresponding to the ID of the respective individual).
## Estimate generation levels for all families.
estimateGenerations(fad)[1:3] ## $`4`
## 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
## 0 0 2 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 3 3 3 2 1
## 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43
## 1 1 1 1 NA NA NA NA NA NA NA NA NA NA NA NA NA NA
##
## $`5`
## 44 45 46 47 48 49 50 51 52 53 54 55
## 0 0 2 2 2 2 2 2 2 2 1 3
## 56 57 58 59 60 61 62 63 64 65 66 67
## 3 3 3 3 3 3 3 3 3 2 2 2
## 68 69 70 71 72 73 74 75 76 77 78 79
## NA 2 1 NA NA NA NA NA NA NA NA 2
## 26050 26051
## NA NA
##
## $`6`
## 80 81 82 83 84 85 86 87 88 89 90 91
## 0 0 2 2 1 1 1 1 1 1 1 1
## 92 93 94 95 96 97 98 99 100 101 102 103
## 1 1 2 2 2 2 2 2 2 3 3 3
## 104 105 106 107 108 109 110 111 112 113 114 115
## 3 2 2 1 1 1 NA NA NA NA 2 NA
## 116 117 118 26052 26053
## NA NA NA 3 3Individuals without generation level (i.e. with an NA) are not connected to
any other individual in the pedigree (and thus most likely represent errors in
the pedigree).
In addition, it is also possible to calculate generation levels relative to a (single) specified individual:
gens <- generationsFrom(fad, id = "4") We can render these generation numbers into the pedigree:
plotPed(fad, family = 4, label2 = gens) 
## Did not plot the following people: 30 31 32 33 34 35 36 37 38 39 40 41 42 432.2 Additional plotting options
If a trait information is available it might be of interest to highlight
affected individuals in the pedigree. Trait information should always be coded
as 0 (or FALSE) for unaffected and 1 (or TRUE) for affected. In the
example below, we use the cancer information from the Minnesota Breast Cancer
Study.
## Extract the cancer trait information.
tcancer <- mbsub$cancer
names(tcancer) <- mbsub$id
## Set the trait.
trait(fad) <- tcancerWe can now extract the trait information from the object or identify directly the phenotyped or affected individuals.
## Extract the trait information.
head(trait(fad))## 1 2 3 4 5 6
## 0 0 0 1 0 0## We can also extract the IDs of the affected individuals.
head(affectedIndividuals(fad))## [1] "4" "11" "37" "54" "84" "122"## Or the IDs of the phenotyped individuals.
head(phenotypedIndividuals(fad)) ## [1] "1" "2" "3" "4" "5" "6"Plotting a FAData object with trait information results in a pedigree plot
with highlighted affected individuals (for kinship2 pedigree plotting:
affected, unaffected and not phenotyped are represented as filled symbols, open
symbols and symbols with a question mark inside, respectively).
## Plotting the pedigree for family "9".
plotPed(fad, family = "9")