This is a short lesson on how to create heatmaps. You will create two heatmaps in this session: one with step-by-step instructions and the other on your own in the exercises.

You are provided two datasets:

  1. One of the average high temperatures for cities in the US

  2. RNA-seq Data from the ENCODE Project

Part One

While there are several ways that you can create heatmaps, we are going to use the heatmap.2 function in the gplots package

library(gplots)

We are going to begin by importing our dataset

Temperatures = read.delim("../dataSets/Temperatures.txt")

We can then preview our dataset:

head(Temperatures)
         X Boston New.York Miami Los.Angeles Aspen
1  January     36       36    74          68    36
2 February     39       40    75          69    39
3    March     45       48    76          70    46
4    April     56       58    79          73    53
5      May     66       68    83          74    63
6     June     76       77    87          78    73
  Anchorage New.Orleans San.Diego San.Francisco Austin
1        22          62        65            57     62
2        26          65        65            60     65
3        34          72        66            62     72
4        44          78        68            63     60
5        56          85        68            64     87
6        63          90        71            67     92
  Phoenix Chicago Honolulu Atlanta Washington.D.C. Seattle
1      67      32       81      51              43      47
2      71      38       81      56              47      51
3      77      47       82      63              56      55
4      85      59       83      72              67      59
5      95      70       85      79              75      65
6     104      80       87      85              84      70
  St..Louis Minneapolis
1        40          24
2        45          29
3        55          41
4        67          58
5        77          69
6        85          79

Notice that the first column of data, should be our row names instead of the first column. To fix this:

row.names(Temperatures) = Temperatures[,1]
Temperatures = Temperatures[,-1]

Now when we preview the dataset, the data frame has the appropriate row names.

head(Temperatures)
         Boston New.York Miami Los.Angeles Aspen Anchorage
January      36       36    74          68    36        22
February     39       40    75          69    39        26
March        45       48    76          70    46        34
April        56       58    79          73    53        44
May          66       68    83          74    63        56
June         76       77    87          78    73        63
         New.Orleans San.Diego San.Francisco Austin Phoenix
January           62        65            57     62      67
February          65        65            60     65      71
March             72        66            62     72      77
April             78        68            63     60      85
May               85        68            64     87      95
June              90        71            67     92     104
         Chicago Honolulu Atlanta Washington.D.C. Seattle
January       32       81      51              43      47
February      38       81      56              47      51
March         47       82      63              56      55
April         59       83      72              67      59
May           70       85      79              75      65
June          80       87      85              84      70
         St..Louis Minneapolis
January         40          24
February        45          29
March           55          41
April           67          58
May             77          69
June            85          79

To plot data in a heatmap you must first transform the data frame into a data matrix. This can be easily done using the data.matrix() function:

TempMatrix=data.matrix(Temperatures)

We are now ready to make our heatmap:

heatmap.2(TempMatrix)

Exercise #1

Now you are going to do the same actions for your ENCODE RNA-seq dataset:

  1. Load the data into a data frame
  2. Clean up the data if needed (make sure you have proper column and row names)
  3. Create a data matrix from the data frame
  4. Plot the data using heatmap.2

Part Two

Congratulations! You have learned how to make a heatmap! However, it looks rather awful. This section is going to cover several options which you can change to improve your heatmap.

Clustering

By default the heatmap.2() function will cluster both the rows and columns. This means that it will group rows and columns that are more similar closer together. Depending on what you are plotting, you may way to remove this option.

To remove column clustering:

heatmap.2(TempMatrix, Colv=NA)

To remove row clustering:

heatmap.2(TempMatrix, Rowv=NA)

To remove all clustering:

heatmap.2(TempMatrix, Colv=NA, Rowv=NA)

Dendrograms

By default the heatmap.2() function will add dendograms to your rows and columns demonstration how they are clustered. If you disable clustering you will also remove the dendrograms. However, you may want to keep the clustering and just remove the dendograms.

To only have row dendrograms:

heatmap.2(TempMatrix, dendrogram="row")