R: converting Ensembl row names to Symbol ID outputs missing values in 'row.names' are not allowed

Question

[user31888]

I have a .csv file as follows:

,TEST1,TEST2
ENSG00000197421,2,0
ENSG00000213753,0,2
ENSG00000168746,0,2
ENSG00000261824,3,0
ENSG00000128310,1,2
ENSG00000235091,9,4

In R, I import the file like this:

 > d <- read.csv("my_file.csv", header=TRUE, row.names=1)
 > d
                TEST1 TEST2
ENSG00000197421     2     0
ENSG00000213753     0     2
ENSG00000168746     0     2
ENSG00000261824     3     0
ENSG00000128310     1     2
ENSG00000235091     9     4

Checking that I do not have any duplicates:

> rownames(d)
[1] "ENSG00000197421" "ENSG00000213753" "ENSG00000168746" "ENSG00000261824"
[5] "ENSG00000128310" "ENSG00000235091"
> colnames(d)
[1] "TEST1" "TEST2"
> any(duplicated(rownames(d)))
[1] FALSE
> any(duplicated(colnames(d)))
[1] FALSE

Load libraries:

> suppressMessages(library("AnnotationDbi"))
> suppressMessages(library("org.Hs.eg.db"))

Then try to convert my Ensembl row names to Symbol in place:

> rownames(d) <- mapIds(org.Hs.eg.db,keys=rownames(d),column="SYMBOL",keytype="ENSEMBL",multiVals="first")
Error in `row.names<-.data.frame`(`*tmp*`, value = value) : 
  missing values in 'row.names' are not allowed

NOTE: Removing the first ',' on 'my_file.csv' did not help neither.

I managed to create a new field with the converted IDs but cannot replace it to the row names:

> d$SYMBOL <- mapIds(org.Hs.eg.db,keys=rownames(d),column="SYMBOL",keytype="ENSEMBL",multiVals="first")
> d
                TEST1 TEST2    SYMBOL
ENSG00000197421     2     0     GGT3P
ENSG00000213753     0     2 CENPBD1P1
ENSG00000168746     0     2 LINC01620
ENSG00000261824     3     0 LINC00662
ENSG00000128310     1     2     GALR3
ENSG00000235091     9     4      <NA>
> d_subset <- subset(d, !is.na(d$SYMBOL))
> d_subset
                TEST1 TEST2    SYMBOL
ENSG00000197421     2     0     GGT3P
ENSG00000213753     0     2 CENPBD1P1
ENSG00000168746     0     2 LINC01620
ENSG00000261824     3     0 LINC00662
ENSG00000128310     1     2     GALR3
> rownames(d) <- d$SYMBOL
Error in `row.names<-.data.frame`(`*tmp*`, value = value) : 
  missing values in 'row.names' are not allowed

I don't get it.

[Charles Plessy]

Here, missing values means NAs, which can not be used as row names. You need to convert them to unique names (because duplicate row names are not allowed).

Answer 1

[seancho]

In your last line, you're still trying to assign rownames(d) <- d$SYMBOL, and not your new d_subset.

rownames(d_subset) <- d_subset$SYMBOL should work.

Alternatively, if you wish to keep all the entries, you could retain the Ensembl names when it is not mapped:

> rownames(d) <- ifelseis.na(d$SYMBOL), rownames(d), d$SYMBOL)

> d
                TEST1 TEST2    SYMBOL
GGT3P               2     0     GGT3P
CENPBD1P1           0     2 CENPBD1P1
LINC01620           0     2 LINC01620
LINC00662           3     0 LINC00662
GALR3               1     2     GALR3
ENSG00000235091     9     4      <NA>

[user31888]

+1 for keeping the ENSG. Thanks !

[user31888]

Missing a bracket. The editor does not want it...

Answer 2

[user31888]

Sorry, but I don't see any missing values in my dataset. And I don't see any duplicates in any field neither. That's what I don't understand.

[Charles Plessy]

ENSG00000235091     9     4      <NA>

Here, <NA> means that you have a missing value.

[user31888]

Actually, I forgot to use the subset of my data frame in my last piece of code

Answer 3

[user31888]

Sorry, but I don't see any miising values in my dataset. That's what I don't understand.

Answer 4

[Liun]

It's because not all your keys (rownames(d)) in org.Hs.eg.db.

> rownames(d)

"ENSG00000197421" "ENSG00000213753" "ENSG00000168746" "ENSG00000261824" "ENSG00000128310" "ENSG00000235091"

> intersect(rownames(d),keys(org.Hs.eg.db,"ENSEMBL"))

"ENSG00000197421" "ENSG00000213753" "ENSG00000168746" "ENSG00000261824" "ENSG00000128310"

if you run this code :

mapIds(org.Hs.eg.db,keys=rownames(d)[1:5],column="SYMBOL",keytype="ENSEMBL",multiVals="first")

it's ok without any error.