Question

I have following type data for human family:

indvidual <- c("John",  "Kris", "Peter",  "King",  "Marry",  "Renu", "Kim",    "Ken", "Lu")
Parent1 <- c(    NA,     NA,     "John",  "John",   "John",    NA,    "Peter",  NA,    NA)
Parent2 <- c(    NA,     NA,    "Kris",   "Kris",  "Renu",   NA,      "Lu",     NA,   NA)
X <-       c(    2,     3,       2,       3,           4,     5,        1.5,      1,    1)
Y <-       c(    3,     3,       2,       2,           2,     3,        1,      3,    2)
pchsize <- c( 4.5,      4.3,     9.2,     6.2,         3.2,   6.4,      2.1,    1.9,  8)
fillcol <- c( 8.5,      8.3,     1.2,     3.2,         8.2,   2.4,      2.6,    6.1,  3.2)
myd <- data.frame (indvidual, Parent1, Parent2, X, Y, pchsize,fillcol)

 indvidual Parent1 Parent2   X Y pchsize fillcol
1      John    <NA>    <NA> 2.0 3     4.5     8.5
2      Kris    <NA>    <NA> 3.0 3     4.3     8.3
3     Peter    John    Kris 2.0 2     9.2     1.2
4      King    John    Kris 3.0 2     6.2     3.2
5     Marry    John    Renu 4.0 2     3.2     8.2
6      Renu    <NA>    <NA> 5.0 3     6.4     2.4
7       Kim   Peter      Lu 1.5 1     2.1     2.6
8       Ken    <NA>    <NA> 1.0 3     1.9     6.1
9        Lu    <NA>    <NA> 1.0 2     8.0     3.2

I want plot something like the following, individuals points are connected to parents (Preferably different line color to Parent1 and Parent2 listed). Also pch size and pch fill is scaled to other variables pchsize and fillcol. Thus plot outline is:

Here is my progress in ggplot2:

require(ggplot2) 
ggplot(data=myd, aes(X, Y,fill = fillcol)) +
  geom_point(aes(size = pchsize, fill = fillcol), pch = "O") +
  geom_text(aes (label = indvidual, vjust=1.25))

Issues unsolved: connecting lines, making size of pch big and fill color at the sametime.

Answer 1

One thing that jumped out to me was to treat this is a network – R has many packages to plot these.

Here’s a very simple solution:
First, I used your parent list to make a sociomatrix – you can generally input networks using edge lists as well – here I put 1 for the first parental relationship and 2 for the second.

psmat <- rbind(c(0, 0, 1, 1, 1, 0, 0, 0, 0),
          c(0, 0, 2, 2, 0, 0, 0, 0, 0),
          c(0, 0, 0, 0, 0, 0, 1, 0, 0),
          rep(0, 9),
          rep(0, 9),
          c(0, 0, 0, 0, 2, 0, 0, 0, 0),
          rep(0, 9),
          rep(0, 9),
          c(0, 0, 0, 0, 0, 0, 2, 0, 0))

Then, using the network package I just hit:

require(network)
plot(network(psmat), coord = cbind(X, Y), vertex.cex = pchsize, 
  vertex.col = fillcol, label = indvidual, edge.col = psmat)

This isn’t terribly pretty in itself, but I think gives you all the basic elements you wanted.

For the colors, I believe the decimal places are just rounded – I wasn’t sure what to do with those.

I know I’ve seen people plot networks in ggplot, so that might give you a better result.

Edit:
So here’s a really messy way of turning your data into a network object directly – someone else might be able to fix it. Additionally, I add an edge attribute (named ‘P’ for parental status) and give the first set a value of 1 and the second set a value of 2. This can be used when plotting to set the colors.

P1 <- match(Parent1, indvidual)
e1 <- cbind(P1, 1:9); e1 <- na.omit(e1); attr(e1, 'na.action') <- NULL
P2 <- match(Parent2, indvidual)
e2 <- cbind(P2, 1:9); e2 <- na.omit(e2); attr(e2, 'na.action') <- NULL

en1 <- network.initialize(9)
add.edges(en1, e1[,1], e1[,2])
set.edge.attribute(en1, 'P', 1)
add.edges(en1, e2[,1], e2[,2], names.eval = 'P', vals.eval = 2)

plot(en1, coord = cbind(X, Y), vertex.cex = pchsize, 
  vertex.col = fillcol, label = indvidual, edge.col = 'P')