Hi,

I have a vector with Ensembl gene IDs

> headgene.id)
[1] "ENSG00000223972" "ENSG00000227232" "ENSG00000278267" "ENSG00000243485"
[5] "ENSG00000274890" "ENSG00000237613"

I am trying to annotate the IDs using biomaRt

> library(biomaRt)

> ensembl <- useMart("ENSEMBL_MART_ENSEMBL",
                   dataset = "hsapiens_gene_ensembl",
                   host = "www.ensembl.org")

However, I get an error. The curious thing is that I get this error only when my vector has length 993 or longer, never when it is shorter, using a random selection of IDs. So this always works:

> mat.cpm.annot <- biomaRt::getBM(attributes = c("ensembl_gene_id", "hgnc_id", "hgnc_symbol", "description"), filter = "ensembl_gene_id", samplegene.id, 992), mart = ensembl, uniqueRows = TRUE)

And this gives me an error:

> mat.cpm.annot <- biomaRt::getBM(attributes = c("ensembl_gene_id", "hgnc_id", "hgnc_symbol", "description"), filter = "ensembl_gene_id", samplegene.id, 993), mart = ensembl, uniqueRows = TRUE)

Error in biomaRt::getBM(attributes = c("ensembl_gene_id", "hgnc_id", "hgnc_symbol",  :
  Query ERROR: caught BioMart::Exception: non-BioMart die():
not well-formed (invalid token) at line 1, column 16292, byte 16292 at /nfs/public/release/ensweb-software/sharedsw/2017_04_03/linuxbrew/Cellar/perl/5.24.1/lib/perl5/site_perl/5.24.1/x86_64-linux-thread-multi/XML/Parser.pm line 187.
XML::Simple called at /nfs/public/release/ensweb/latest/live/mart/www_90/biomart-perl/lib/BioMart/Query.pm line 1935.

> sessionInfo()
R version 3.4.2 (2017-09-28)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: CentOS release 6.5 (Final)

Matrix products: default
BLAS: /share/apps/cto/packages/R/3.4.2/lib64/R/lib/libRblas.so
LAPACK: /share/apps/cto/packages/R/3.4.2/lib64/R/lib/libRlapack.so

locale:
 [1] LC_CTYPE=en_GB.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_GB.UTF-8        LC_COLLATE=en_GB.UTF-8    
 [5] LC_MONETARY=en_GB.UTF-8    LC_MESSAGES=en_GB.UTF-8   
 [7] LC_PAPER=en_GB.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_GB.UTF-8 LC_IDENTIFICATION=C       

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] biomaRt_2.32.1

loaded via a namespace (and not attached):
 [1] Rcpp_0.12.13         IRanges_2.10.5       XML_3.98-1.9        
 [4] digest_0.6.12        bitops_1.0-6         DBI_0.7             
 [7] stats4_3.4.2         RSQLite_2.0          rlang_0.1.2         
[10] blob_1.1.0           S4Vectors_0.14.7     tools_3.4.2         
[13] bit64_0.9-7          Biobase_2.36.2       RCurl_1.95-4.8      
[16] bit_1.1-12           parallel_3.4.2       compiler_3.4.2      
[19] BiocGenerics_0.22.1  AnnotationDbi_1.38.2 memoise_1.1.0       
[22] tibble_1.3.4        
 

Any idea, what is going on?

Cheers,

Georg

That's a new error to me!  I suspect that something is wrong with the back end database, rather than with the biomaRt package.

One thing you can try is to use one of the mirror services, to see if that is unaffected, e.g:

ensembl <- useMart("ENSEMBL_MART_ENSEMBL",
                   dataset = "hsapiens_gene_ensembl",
                   host = "asia.ensembl.org")

Alternatively, you can try the developmental version of biomaRt.  It's not recommended to run queries with more than 500 search values, and although in practice it's often fine occasionally results won't be returned, but you'll have no idea that's happened.  The devel package has a modification that breaks your query down into chunks of 500 and runs the independently and then splices the results back together.  Since your issue seems so deterministic perhaps this modification will be sufficient.  You can install using:

BiocInstaller::biocLite('grimbough/biomaRt')

A quick test for me suggests the uniqueRows argument won't work properly at the moment, but you can do it in post processing yourself.

I can confirm upgrading bioconductor to version 3.6. solved the problem.