Question

Adding Enrichresults for creating plot

0

Entering edit mode

felix.thaler • 0

@felixthaler-23378

Last seen 4.6 years ago

I am performing pathway analysis using the ReactomePA package and the clusterProfiler package. I did the pathway analysis and got 2 objects, one by using the enrichPathway() function from ReactomePA and one by using the enrichKEGG() function from clusterProfiler.

Both objects have the same class: enrichResult.

Now my question: I am able to do plots which each object itself, but can I also make a plot that shows the pathways of both objects? Or is there another way I can achieve this?

Thanks a lot for your help, I really like both packages!

ReactomePA clusterProfiler • 2.9k views

ADD COMMENT • link 4.6 years ago felix.thaler • 0

score 0 · Answer 1 · 2020-04-19

0

Entering edit mode

Guangchuang Yu ★ 1.2k

@guangchuang-yu-5419

Last seen 7 days ago

China/Guangzhou/Southern Medical Univer…

you can use the mergeresult function that is documented in https://github.com/GuangchuangYu/enrichment4GTExclusterProfiler.

ADD COMMENT • link 4.6 years ago Guangchuang Yu ★ 1.2k

0

Entering edit mode

Thanks, but the link gives me the 404 Error message.

how should i use the merge_result function? I read the clusterProfiler reference manual

simply putting in the list of enrichResults i want to merge doesn't work unfortunately....

ADD REPLY • link 4.6 years ago felix.thaler • 0

0

Entering edit mode

Somehow the underscore in the URL malformatted it... the link above should be:

https://github.com/GuangchuangYu/enrichment4GTEx_clusterProfiler/

merge_results expects multiple lists!

The code below works (inspired by your use case):

library(clusterProfiler)
library(ReactomePA)

# selects some (100) genes for overrepresentation analysis
data(geneList, package='DOSE')
de <- names(geneList)[1:100]

# note high cutoff p-value!
# did this to make sure some pathways were included
y.reactome = enrichPathway(de, pvalueCutoff=0.5)
y.kegg <- enrichKEGG(gene, pvalueCutoff=0.5)

# merge the results
out <-  merge_result( list("Reactome"=y.reactome, "KEGG"=y.kegg))

# generate a dotplot from merged results.
# note that in the example below 5 (only) pathways will be shown.
# Since by definition there is no overlap in enriched pathways,
# because the sources are different (!) (Reactome vs KEGG),
# five pathways per pathway database will be shown.
#  
# The number below each column name are the number of genes
# that could be annotated in each database.
dotplot(out, showCategory=5)

ADD REPLY • link 4.6 years ago Guido Hooiveld ★ 4.1k

0

Entering edit mode

It worked, but i got only 17 merged results and 16 plots even though i expected 39. The console put out the following:

--> No gene can be mapped....
--> Expected input gene ID: 51144,6476,54659,3420,130,47
--> return NULL...
--> No gene can be mapped....
--> Expected input gene ID: 9972,3671,4860,2548,7873,8635
--> return NULL...
--> No gene can be mapped....
--> Expected input gene ID: 3418,6476,226,8789,221823,51
--> return NULL...
Fehler in `[.data.frame`(result, , c("ID", "Description")) : 
  undefined columns selected

Thanks for your help!

ADD REPLY • link 4.6 years ago felix.thaler • 0

0

Entering edit mode

Please post your full code, ideally fully reproducible (thus using the included data set).... Your output as such is not very informative....

ADD REPLY • link 4.6 years ago Guido Hooiveld ★ 4.1k

0

Entering edit mode

Alright, that's my code

merged_pathways <- list()
all_plots <- list()

# entrez_gene_ids is a  list of 39 character vectors containing Entrez gene IDs

for (file in entrez_gene_ids) {

  pathway_reactome <- enrichPathway(gene = file, 
                                    # not using universe argument at the moment
                                    organism = "human",
                                    pvalueCutoff = 0.05,
                                    pAdjustMethod = "BH",
                                    qvalueCutoff = 0.2)

  pathway_kegg <- enrichKEGG(gene = file, 
                             # not using the universe argument at the moment
                             organism = "hsa",
                             pAdjustMethod = "BH",
                             pvalueCutoff = 0.05,
                             qvalueCutoff = 0.2)

  merged_result <- merge_result( list("Reactome"=pathway_reactome, "KEGG"=pathway_kegg))

  merged_pathways[[length(merged_pathways)+1]] <- merged_result 

  merged_plot <- dotplot(merged_result, showCategory = 15)

  all_plots[[length(all_plots)+1]] <- merged_plot

 }

ADD REPLY • link 4.6 years ago felix.thaler • 0

0

Entering edit mode

I haven't fully checked your code, but based on the output you showed earlier: could it be that some of your input vectors consist of entrez ids that are not HUMAN entrez ids, or only of ids that are not (yet) annotated to a Reactome or KEGG pathway? That would explain that 'no gene cane be mapped'....

ADD REPLY • link 4.6 years ago Guido Hooiveld ★ 4.1k

0

Entering edit mode

They're all human genes... when i did the enrichment analysis without merging both results it worked, so maybe the Error Message is related to the merge_result function...

ADD REPLY • link 4.6 years ago felix.thaler • 0