Question

clusterProfiler: exclude pathways and add gene names to cnetplot

0

Entering edit mode

Mr.RB ▴ 20

@mrrb-15633

Last seen 3.1 years ago

Package: clusterProfiler by Guangchuang Yu

Let start with the example code (just the same as from the vignette):

library(clusterProfiler)
library(enrichplot)

# basic analysis
data(geneList, package="DOSE")
kegg <- gseKEGG(geneList = geneList, organism = 'hsa')

# plotting
cnetplot(kegg, categorySize="pvalue", foldChange=geneList)

I want to change two things in this cnetplot

Exclude pathways in the graph
Providing gene names instead of the IDs

1. Using the example:

require(dplyr)
head(kegg@result) %>% select(Description, setSize)

                               Description setSize
hsa04510                    Focal adhesion     188
hsa00982 Drug metabolism - cytochrome P450      49
hsa03050                        Proteasome      42
hsa03030                   DNA replication      33
hsa05340          Primary immunodeficiency      34
hsa04151        PI3K-Akt signaling pathway     320

Let's say I do not want the genes for the Focal adhesion in the cnet graph, how can this be accomplished.

2. using the example

Instead of the ids (such as 6891) I want to show the gene names which I provide myself (or maybe this can be done automatically)

clusterprofiler pathway analysis pathways • 3.0k views

ADD COMMENT • link updated 4.0 years ago by Guangchuang Yu ★ 1.2k • written 4.0 years ago by Mr.RB ▴ 20

score 0 · Answer 1 · 2018-05-28

0

Entering edit mode

Guangchuang Yu ★ 1.2k

@guangchuang-yu-5419

Last seen 5 weeks ago

China/Guangzhou/Southern Medical Univer…

1. same question posted on https://github.com/GuangchuangYu/DOSE/issues/20#issuecomment-391802809.

please upgrade enrichplot to v >= 1.0.1, and then you can pass selected Pathway Names (selected from the Description column) to showCategory parameter.

2. you can use DOSE::setReadable to translate gene ID to gene symbol.

ADD COMMENT • link 4.0 years ago Guangchuang Yu ★ 1.2k

0

Entering edit mode

Dear Guangchuang Yu,

Regarding question 1:

The version I currently have is enrichplot_1.1.1.001 , which is >=1.0.1 as recommended by you. However when I pass a vector of pathway names to the cnetplot command ( like so x <- cnetplot(kegg, showCategory = kegg$Description[-2]) ) it will give this error:

Error in 1:n : NA/NaN argument
In addition: Warning messages:
1: In 1:n : numerical expression has 38 elements: only the first used
2: In V(g)$size[1:n] <- size : NAs introduced by coercion

Can't figure out what I'm doing wrong here.

Regarding question 2:

After re-reading this question I think I wasn't that clear with what I mean. I love that function, however what I meant is that I could pass this to cnetplot. Of course I could convert the ids to the gene names before calling the gseKEGG function, but this will probably interfere with mapping. Therefore, I want to be able to pass the gene names to cnetplot

ADD REPLY • link 4.0 years ago Mr.RB ▴ 20

0

Entering edit mode

pls use enrichplot v >= 1.0.2 or v >= 1.1.2, which address the issue you reported.
you misunderstand how setReadable works.


## run gseKEGG as usual

kk <- gseKEGG()

## convert gene ID to symbol

kk2 <- setReadable(kk) # refer to man for additional parameters

## plot

cnetplot(kk2)

ADD REPLY • link 4.0 years ago Guangchuang Yu ★ 1.2k

0

Entering edit mode

1. Thankyou! that version seems to work fine

2. The setReadable will of course only work whenever a db is available for the species in question, this is not the case for me. Therefore I mentioned gene names which I provide myself in my question I looked at the code of and thought I could easily mimic the setReadable function like so:

kegg@gene2Symbol <- as.character(gene.name.vector)
kegg@keytype <- 'auto'
kegg@result <- kegg@result
kegg@readable <- TRUE

However this will still only show the ids (and loses coloring in the network), instead the names I provided in the gene.name.vector, is there any way to provide these manually?

ADD REPLY • link 4.0 years ago Mr.RB ▴ 20