topGo Error in .local(.Object, ...) : allGenes must be a factor with 2 levels

Question

[Ric]

Hi, I have got an Error in .local(.Object, ...) : allGenes must be a factor with 2 levels using the following code

> library(edgeR)
> library(topGO)
...
> tr <- glmTreat(fit, contrast=B.LvsP, lfc=log2(1.5))
> topTags(tr)
Coefficient:  -1*Leaves.2 1*Leaves.3 
          logFC unshrunk.logFC logCPM  PValue     FDR
sp0090975   2.6            2.6    2.8 5.7e-14 2.7e-09
sp0037632  -3.0           -3.0    3.4 1.7e-13 2.7e-09
sp0074153  -3.9           -3.9    3.8 1.8e-13 2.7e-09
sp0008306   3.2            3.2    2.9 1.8e-13 2.7e-09
sp0073530  -4.5           -4.5    3.4 2.3e-12 2.3e-08
sp0025713  -3.9           -3.9    4.4 2.6e-12 2.3e-08
sp0037721   7.8            8.0    2.4 2.7e-12 2.3e-08
sp0083660   2.0            2.0    4.4 3.2e-12 2.3e-08
sp0052245  -2.9           -2.9    4.9 3.9e-12 2.6e-08
sp0071520  -3.3           -3.3    2.7 5.8e-12 3.4e-08


> geneID2GO <- readMappings(file = "~GOmapping.tsv")
> head(geneID2GO)
$`"V1"`
[1] "\"V14\""

$`"sp0000005"`
[1] "\"GO:0003723\""

$`"sp0000006"`
[1] "\"GO:0016021\""

$`"sp0000007"`
[1] "\"GO:0003700" "GO:0006355"   "GO:0043565\""

$`"sp0000016"`
[1] "\"GO:0046983\""

$`"sp0000017"`
[1] "\"GO:0004672" "GO:0005524"   "GO:0006468\""

> geneUniverse <- names(geneID2GO)
> head(geneUniverse)
[1] "\"V1\""        "\"sp0000005\"" "\"sp0000006\"" "\"sp0000007\"" "\"sp0000016\""
[6] "\"sp0000017\""

> genesOfInterest <- rownames(tr@.Data[[1]])
> head(genesOfInterest)
[1] "sp0025247" "sp0025250" "sp0025268" "sp0025270" "sp0025282" "sp0056834"

> geneList <- factor(as.integer(geneUniverse %in% genesOfInterest))

> head(geneList)
[1] 0 0 0 0 0 0
Levels: 0

> names(geneList) <- geneUniverse
> myGOdata <- new("topGOdata", 
+                 description="My project", 
+                 ontology="BP", 
+                 allGenes=geneList,  
+                 annot = annFUN.gene2GO, 
+                 gene2GO = geneID2GO)
Error in .local(.Object, ...) : allGenes must be a factor with 2 levels

What did I miss?

Thank you in advance.

[Kevin Blighe]

Are you sure that the line geneList <- factor(as.integer(geneUniverse %in% genesOfInterest)) is doing what you expect it to do? First converting to integers and then factors doesn't look right to me. I'm neither sure that the lookup with %in% is going to work as you'd expect. The %in% operator will just return indices, but not actual values.

Your geneList variable appears to be just zeros, i.e., a single factor.

Happy to assist further

Answer 1

[Ric]

> genesOfInterest <- rownames(tr@.Data[[1]])
> head(genesOfInterest)
[1] "sp0025247" "sp0025250" "sp0025268" "sp0025270" "sp0025282" "sp0056834"
> geneList <- factor(as.integer(grep(genesOfInterest, geneUniverse)))
Warning message:
In grep(genesOfInterest, geneUniverse) :
  argument 'pattern' has length > 1 and only the first element will be used
> head(geneList)
factor(0)
Levels: 
> names(geneList) <- geneUniverse
Error in names(geneList) <- geneUniverse : 
  'names' attribute [44620] must be the same length as the vector [0]

Your second option geneUniverse <- gsub("\\\"", "", geneUniverse)) will be used with grep together?

[Kevin Blighe]

Indeed, apologies, grep doesn't work with multiple values.

The grep code should be:

indices <- c()

for (i in 1:length(genesOfInterest)) indices <- c(indices, grep(genesOfInterest[i], geneUniverse))

geneList <- factor(as.integer(indices))

The second option should work without grep.

[Ric]

I have got 'names' attribute [44620] must be the same length as the vector [33975]

> genesOfInterest <- rownames(tr@.Data[[1]])
> head(genesOfInterest)
[1] "sp0025247" "sp0025250" "sp0025268" "sp0025270" "sp0025282" "sp0056834"
> indices <- c()
> for (i in 1:length(genesOfInterest)) indices <- c(indices, grep(genesOfInterest[i], geneUniverse))
> geneList <- factor(as.integer(indices))
> head(geneList)
[1] 12064 12076 26933 18988 18992 18994
33975 Levels: 2 3 4 5 6 8 10 11 12 13 14 15 18 19 22 23 24 25 26 28 29 30 31 32 33 34 35 ... 44618
> names(geneList) <- geneUniverse
Error in names(geneList) <- geneUniverse : 
  'names' attribute [44620] must be the same length as the vector [33975]

[Kevin Blighe]

I think that your command > names(geneList) <- geneUniverse should be

names(geneList) <- geneUniverse[indices]

Nevertheless, in your other thread, which appears to be a duplicate of this, didn't you say that the geneList vector needs to have P values and then names as the gene names?

[Ric]

Thank you for your help. Yes, I found another tutorial here which took a slightly different approach.

Answer 2

[Ric]

I have followed this tutorial for topGO. Only thing which I changed was:

> genesOfInterest <- read.table("interestinggenes.txt",header=FALSE)
> genesOfInterest <- as.character(genesOfInterest$V1)

to

> genesOfInterest <- rownames(tr@.Data[[1]])

Do you have any idea how to fix the geneList?

Thank you in advance.

[Kevin Blighe]

I see, so that function is indeed just looking for indices (and not values).

It just looks like your matching is not working.

For example, "\"sp0000005\"" is not the same as "sp0000005"

a <- c("\"sp0000005\"")
b <- c("sp0000005")

a %in% b
[1] FALSE

b %in% a
[1] FALSE

...however, grep returns indices too and only looks for partial matches (which can be risky at times):

grep(b, a)
[1] 1

[Ric]

I found another tutorial here but I do not know how to use it only in R

[Kevin Blighe]

Dear Ric,

I think that you just need to change the following line

geneList <- factor(as.integer(geneUniverse %in% genesOfInterest))

to:

EDIT: see next answer thread below

The other option is to add a new line of code after geneUniverse <- names(geneID2GO) that ensures that your variables in geneUniverse and genesOfInterest have the same name:

geneUniverse <- gsub("\\\"", "", geneUniverse))