I used the bitr function from the clusterProfiler package to convert gene symbols from a DE experiment to UniProt protein ids. For some unique gene symbols, there are multiple UniProt ids.
Surely each gene id should map to a single protein and each protein has a unique id. So is my code correct and does it matter that there are multiple UniProt ids for a single gene?
My code is
Genes <- c("AACS", "ACAA2", "ACADM", "ACLY", "ACOT8")
Protein_IDs <- bitr(Genes, fromType="SYMBOL", toType="UNIPROT", OrgDb="org.Hs.eg.db") # returns 15 rows
test <- distinct(Protein_IDs, UNIPROT, .keep_all = TRUE) # returns 15 rows
SYMBOL UNIPROT
AACS Q86V21
AACS A0A024RBV2
ACAA2 B3KNP8
ACAA2 P42765
ACADM A0A0S2Z366
ACADM P11310
ACADM B7Z9I1
ACADM Q5HYG7
ACADM Q5T4U5
ACADM B4DJE7
ACLY A0A024R1T9
ACLY P53396
ACLY Q4LE36
ACLY A0A024R1Y2
ACOT8 O14734
•
written
by
peter.berry5 •
2 answers
I used the following package and code to get the uniprot ids and names of the proteins which my DE genes code for.
Library (mygene)
Genes <- c("AACS", "ACAA2", "ACADM", "ACLY", "ACOT8")
Protein_IDs <- queryMany(Genes, scopes = "symbol",
fields = c("name", "uniprot", "ensemblgene"),
species = "human", as_dataframe = "True")
and then used
df1 <- data.frame(hmap_4$query)
df2 <- data.frame(hmap_4$name)
df3 <- data.frame(hmap_4$uniprot.Swiss.Prot)
to extract the info I wanted.
However, today the last line gave the following error
Error in (function (..., row.names = NULL, check.rows = FALSE, check.names = TRUE, :
arguments imply differing number of rows: 0, 1, 2
When I looked at the structure of Protein IDs (which has class DFrame) I noticed that while "query" and "name" are both of type character with 615 onservations, "uniprot.Swiss.Prot" is of type list with 615 entries.
Can anybody advise how I extract this list? Thanks
apologies, hmap_4 is a typo from a different analysis. the lines refering to df should read
df1 <- data.frame(Protein_IDs$query)
df2 <- data.frame(Protein_IDs$name)
df3 <- data.frame(Protein_IDs$uniprot.Swiss.Prot)
str(Protein_IDs)
gives
Formal class 'DFrame' [package "S4Vectors"] with 6 slots
..@ rownames : NULL
..@ nrows : int 443
@ listData :List of 7
.. ..$ query : chr [1:443] "AACS" "ABCB6" "ABCF1" "ABT1" ...
.. ..$ _id : chr [1:443] "65985" "10058" "23" "29777" ...
.. ..$ X_score : num [1:443] 87.7 90.3 87.9 88.4 87.1 ...
.. ..$ name : chr [1:443] "acetoacetyl-CoA synthetase" "ATP binding cassette subfamily B member 6 (Langereis blood group)" "ATP binding cassette subfamily F member 1" "activator of basal transcription 1" ...
.. ..$ notfound : logi [1:443] NA NA NA NA NA NA ...
.. ..$ uniprot.Swiss.Prot:List of 443
.. .. ..$ : chr "Q86V21"
.. .. ..$ : chr "Q9NP58"
etc.
I tried
Protein_IDs["uniprot.Swiss.Prot"]
but didn't quite get the solution I wanted. However, your explanation of DFrame which I hadn't encountered before lead me to the following code
Symbol <- c("AACS", "ACAA2", "ACADM", "ACLY", "ACOT8")
FC <- c("1", "1.5", "-.2", "6", "-10")
Gene <- data.frame(Symbol, FC)
names(Gene)[1] <- "HGNC_Symbol"
Protein.IDs <- queryMany(Gene$HGNC_Symbol, scopes = "symbol",
fields = c("name", "uniprot", "ensemblgene"),
species = "human", as_data.frame = "True")
df3 <- Protein.IDs["uniprot.Swiss.Prot"]
df4 <- as.data.frame(matrix(unlist(Protein.IDs),
nrow=length(unlist(Protein.IDs[1]))))
df4 <- dplyr::select(df4, c(V1, V4, V5))
names(df4)[1] <- "HGNC_Symbol"
names(df4)[2] <- "Protein"
names(df4)[3] <- "uniprot.ID"
df5 <- dplyr::full_join(df4, Gene, by.x = "HGNC_Symbol",
by.y = "HGNC_Symbol")
which now gives the following error
Joining, by = "HGNC_Symbol"
Error: Can't join on `x$HGNC_Symbol` x `y$HGNC_Symbol` because of incompatible types.
i `x$HGNC_Symbol` is of type <list>>.
i `y$HGNC_Symbol` is of type <character>>.
creating df$5 so I have a data frame with Gene symbol, FC, protein name and uniprot ID is the ultimate goal.
> library(mygene)
> library(dplyr)
> HGNC_Symbol <- c("AACS", "ACAA2", "ACADM", "ACLY", "ACOT8")
> FC <- c("1", "1.5", "-0.2", "6", "-10")
> df1 <- data.frame(HGNC_Symbol, FC)
> Protein.IDs <- queryMany(df1$HGNC_Symbol, scopes = "symbol",
+ fields = c("name", "uniprot", "ensemblgene"),
+ species = "human", as_data.frame = "True")
Finished
> df2=as.data.frame(Protein.IDs[c("query","name","uniprot.Swiss.Prot")])
> df1 %>%
+ inner_join(df2, by=c("HGNC_Symbol"="query"))
HGNC_Symbol FC name uniprot.Swiss.Prot
1 AACS 1 acetoacetyl-CoA synthetase Q86V21
2 ACAA2 1.5 acetyl-CoA acyltransferase 2 P42765
3 ACADM -0.2 acyl-CoA dehydrogenase medium chain P11310
4 ACLY 6 ATP citrate lyase P53396
5 ACOT8 -10 acyl-CoA thioesterase 8 O14734
There seems to be discrepancy between outputs I am getting and your posting. Please check your R and R library versions.
You could also use the UniProt IDmapping service at https://www.uniprot.org/uploadlists . It is possible to map to UniProtKB/Swiss-Prot only, i.e. not have unreviewed entries returned. This can also be done programmatically as described at https://www.uniprot.org/help/api%5Fidmapping
Log in to answer this question.
This seems to use a really lenient mapping with unreviewed entries etc. You may have better luck using biomaRt.
HUGO entry for
ACADMindeed lists only one UniProt accession.You can download an official list of human gene symbols and their corresponding UniProt ID's from HUGO site using a custom download. Select things you want in output.
Apologies. I actually wanted to report a solution I found using the "mygene" package. I hit the delete button in error.
Edit:
see below for an update and further query
Please edit your answer and add some code so people facing similar problems will have a starting point for their solutions.