Annotating Proteins in heatmap with 1 or more pathways using R.

Question

[adam08]

Hello,

I relatively new to programming with R and I was wondering if someone could help me in providing suggestions on annotating proteins in a heatmap. Here I have tried the following code, but the heatmap generated is not color coded correctly. I am also unsure how to add multiple annotations to a single protein. E.g. if I wanted to annotate Protein 4 with all 3 paths or with 2 specific paths how could I do this? Any help would be greatly appreciated.

# Read the CSV file
cultureconditions <- read.csv("Sample proteomic data3.csv", row.names = 1)

# These are the names of the columns 
new_column_names <- c("condition1", "condition2", "condition3", "condition4")  
colnames(cultureconditions) <- new_column_names

#Generate annotation for rows
annotation_row = data.frame(
    GenePaths = factor(rep(c("Path1", "Path2", "Path3"), c(17, 5, 18)))
)

#Color code annotations
ann_colors = list(GenePaths = c(Path1 = "#7570B3", Path2 = "#E7298A", Path3 = "#66A61E")
)

# Create the heatmap
# scale=row allows for normalization with z scores 
#cluster_cols prevents the columns from clustering 

heatmap <- pheatmap(cultureconditions, scale = "row", cluster_row = FALSE, cluster_cols = FALSE, annotation_row = annotation_row, 
    annotation_colors = ann_colors, angle_col = 45, fontsize = 16)

This is the result I get:

[Mark]

When I run your code with a dummy dataset, it produces a figure as expected with Genepaths colored. There's something with your dataset. Could you post the first 20 rows of your dataset? An easy way to do this is to deput like this: dput(cultureconditions[1:20,]). If you can not post your dataset, then create a dummy one (see pheatmap docs for an example)

[adam08]

Thank you for your response! I have attached the code results from the code below.

> dput(cultureconditions[1:20,])

structure(list(condition1 = c(153.6286609, 8495.577956, 1283.506874, 
5376.004328, 1345, 1263.966509, 780.2137199, 3096.534091, 5710.099775, 
437.9382486, 288.10209, 755.1863589, 1160.239461, 241.3590358, 
17675.16009, 9146.903437, 89.29736361, 700, 563.0639983, 4038.300533
), condition2 = c(130, 7304.219269, 1192.080765, 4922.022244, 
2235.794576, 1203.40006, 1291.896364, 2880.33954, 5002.754073, 
352.769094, 200, 500, 1108.729595, 270.5404985, 19282.83952, 
8766.93408, 121.0509699, 900, 588.2291337, 4067.275171), condition3 = c(563.7867629, 
29974.60495, 2494.897555, 8144.586067, 16165.33354, 4376.277767, 
2110.413052, 6043.137715, 8000, 1200, 800, 1206.256272, 2688.307957, 
811.9269546, 28301.27059, 14193.071, 221.0048808, 4931.831031, 
1243.234106, 8258.032841), condition4 = c(700, 30000, 2100, 7006.0668, 
5000, 2658.375157, 1185.035407, 2855.958155, 3958.803261, 768.209857, 
700, 856.2116643, 881.3946047, 1134.283766, 26172.3203, 14865.08983, 
197.3119048, 3000, 1234, 5624.147258)), row.names = c("Protein 1", 
"Protein 2", "Protein 3", "Protein 4", "Protein 5", "Protein 6", 
"Protein 7", "Protein 8", "Protein 9", "Protein 10", "Protein 11", 
"Protein 12", "Protein 13", "Protein 14", "Protein 15", "Protein 16", 
"Protein 17", "Protein 18", "Protein 19", "Protein 20"), class = "data.frame")

Appreciate your time in effort in helping me with this!

[Mark]

Thanks for posting the data. See my answer.

Answer 1

[Mark]

Your annotation_row dataframe needs to have the rows named. This is how pheatmap matches which rows to annotate. Note your row names in your original dataframe need to match exactly the row names in the annotation_row. So Protein1 and Protein 1 are different.

annotation_row = data.frame(
    GeneClass = factor(rep(c("Path1", "Path2", "Path3"), c(10, 4, 6)))
)
rownames(annotation_row) = paste("Protein ", 1:20, sep = "")

I also noticed your annotation_row has the wrong dimensions. It should be 20 rows like your first dataframe.

To improve your code I suggest these edits:

Instead of:

new_column_names <- c("condition1", "condition2", "condition3", "condition4")  
colnames(cultureconditions) <- new_column_names

Just do:

colnames(cultureconditions) <- c("condition1", "condition2", "condition3", "condition4")

[adam08]

Thank you it worked out very nicely! Appreciate all your help. Would you happen to know how I could add multiple annotations to a single protein? e.g I would add to Protein 4 an annotation for all 3 paths or for another protein an annotation of 2 paths?

[Mark]

Add another item to annotation_row and ann_colors eg annotation_row = data.frame( GenePaths = factor(rep(c("Path1", "Path2", "Path3"), c(17, 5, 18))), ProteinPaths = factor(rep(c("Path1", "Path2", "Path3"), c(17, 5, 18))) )

See the docs they have an example of this.

[adam08]

Thank you very much again! I will give it a try. Also, my apologies but could clarify which pheatmap doc you are referring to? This has been my reference for pheatmaps currently: https://jokergoo.github.io/ComplexHeatmap-reference/book/integrate-with-other-packages.html

[Mark]

The reference manual has a huge tutorial. See page 5 https://cran.r-project.org/web/packages/pheatmap/pheatmap.pdf

[adam08]

Thank you very much! I did look through this doc. I think the trouble that I am having is that I would like to use the same data frame (GenePaths) to annotate multiple proteins. In other words, I would like to annotate select proteins to suggest that they are a part of Path1 and Path2 etc. Would you happen to know if I can do this with the same data frame?

[Mark]

There's two ways to do this. I would recommend the what I suggested above about creating another item in the dataframe. Another way is to add another factor to your items to indicate shared paths. eg: factor(rep(c("Path1", "Path2", "Path3", "Path1.Path3"), c(17, 5, 18, 3))), here "Path1.Path3" indicate a shared path.

Don't forget to create a color for the extra items you create.

[adam08]

Thank you very much! Apologies for the repeated questions. I tried the second option that you suggested but it gave the following error shown below.

Error in .rowNamesDF<-(x, value = value) : invalid 'row.names' length

Here is my code below:

  colnames(cultureconditions) <- c("condition1", "condition2", "condition3", "condition4")

#Generate annotation for rows
annotation_row = data.frame(
    GenePaths = factor(rep(c("Path1", "Path2", "Path3","Path1.Path3"), c(17, 5, 18,3)))
)
rownames(annotation_row) = paste("Protein ", 1:40, sep = "")

#Color code annotations
ann_colors = list(GenePaths = c(Path1 = "#7570B3", Path2 = "#E7298A", Path3 = "#66A61E", Path1.Path2 = "#38B41E")
)

# Create the heatmap

heatmap <- pheatmap(cultureconditions, scale = "row", cluster_row = FALSE, cluster_cols = FALSE, annotation_row = annotation_row, 
    annotation_colors = ann_colors, angle_col = 45, fontsize = 16)  

It seems that my rows for my "culturecondition" and annotation.rows are not matching as it provides me with this:

> dim(cultureconditions)
[1] 40  4
> dim(annotation_row)
[1] 43  1

I am not sure how to provide these additional row names given that it is the same row that I am trying annotate a second time.

[Mark]

annotation_row has 43 rows but you are trying to rename 40. Either increase this 1:40 to 1:43 or fix the number of rows you have.

You already knew the problem, the number of items/rows do not match. Make them match.