Module 1: Bonus Exercise Results

Load packages and data, subset needed columns:

library(MASS)
data(crabs)

Learn more about the data:

?crabs
head(crabs)

##   sp sex index   FL  RW   CL
## 1  B   M     1  8.1 6.7 16.1
## 2  B   M     2  8.8 7.7 18.1
## 3  B   M     3  9.2 7.8 19.0
## 4  B   M     4  9.6 7.9 20.1
## 5  B   M     5  9.8 8.0 20.3
## 6  B   M     6 10.8 9.0 23.0
##     CW  BD
## 1 19.0 7.0
## 2 20.8 7.4
## 3 22.4 7.7
## 4 23.1 8.2
## 5 23.0 8.2
## 6 26.5 9.8

Subset needed columns:

crabs_meas <-  crabs[,c("FL","RW","CL","CW","BD")]

Perform hierarchical clustering:

c_dist <- dist(crabs_meas)
c_hclust <- hclust(c_dist)
plot(c_hclust)

Colour-code samples based on cluster assignment. Assume there are two clusters.

c_clusters = cutree(c_hclust,k = 2)

Now create a pairs plot, but colour-code by: 1. by gene-expression based clusters 2. by species 3. by sex

pairs(
    crabs_meas, 
    col = c("orchid","forestgreen")[c_clusters]
)

pairs(
    crabs_meas, 
    col = c("orchid","forestgreen")[factor(crabs$sp)]
)

pairs(
    crabs_meas, 
    col = c("orchid","forestgreen")[factor(crabs$sex)]
)

Hierarchical clustering:

h <- hclust(dist(crabs_meas),method="ward.D2")
c2 <- cutree(h,k=2)

hclust_fun <- function(x){
    f <- hclust(x, method = "ward.D2");
    return(f)
}

library(RColorBrewer)
heatmap(
    as.matrix(crabs_meas),
    hclustfun = hclust_fun,
    col = brewer.pal("Blues",n=8),
    RowSideColors = c("pink","brown")[c2], 
    ColSideColors = rep("green",5)
)

Plot by sex:

heatmap(
    as.matrix(crabs_meas),
    hclustfun = hclust_fun,
    col = brewer.pal("Blues",n=8),
    RowSideColors = c("pink","brown")[factor(crabs$sex)], 
    ColSideColors = rep("green",5)
)