I'm trying to get gene names for the chromosomal location of SNPs (i.e. not trying to gene expression, etc. with SNP, literally just which gene the SNP falls in) using the dbSNP IDs (the ones that start with "rs"). When I supply a long list of attributes to return, I get no results (for this arbitrary SNP), but when I ask only for ENSG gene ID, I get a result. I've looked through the documentation for a flag to change, but have been unable to find one. Thanks!

mart.snps <- useMart('ENSEMBL_MART_SNP', 'hsapiens_snp')

snp = "rs368991480"

getBM(attributes = c("refsnp_id", "ensembl_gene_stable_id", "associated_gene", "chr_name", "chrom_start", "chrom_end", "chrom_strand"),
      filters = "snp_filter",
      values = snp,
      mart = mart.snps)

# returns 0 rows

getBM(attributes = c("refsnp_id", "ensembl_gene_stable_id"),
      filters = "snp_filter",
      values = snp,
      mart = mart.snps)

# returns 1 row

> sessionInfo()
R version 3.3.2 (2016-10-31)
Platform: x86_64-apple-darwin13.4.0 (64-bit)
Running under: OS X Yosemite 10.10.5

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

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

other attached packages:
[1] biomaRt_2.30.0

loaded via a namespace (and not attached):
 [1] IRanges_2.8.1        parallel_3.3.2       DBI_0.5-1            tools_3.3.2          RCurl_1.95-4.8       memoise_1.0.0        Rcpp_0.12.8         
 [8] Biobase_2.34.0       AnnotationDbi_1.36.0 RSQLite_1.1-1        S4Vectors_0.12.1     BiocGenerics_0.20.0  digest_0.6.11        stats4_3.3.2        
[15] bitops_1.0-6         XML_3.98-1.5        

A good rule of thumb is to try doing a similar query at the Ensembl Biomart site if it fails within biomaRt. When I try that, I can make a query for everything but the associated gene (same as in biomaRt). But when I add in the associated_gene (which if you check listAttributes is the 'Associated gene with phenotype' filter), this pops up an extra dataset that I then have to choose. Since we are now using two datasets, this implies we are doing a join across two different databases.

Since biomaRt is designed to do directed queries against a single database, it appears to fail when asked to do a join across different databases. In addition, when I try to do the query including associated_gene at the Ensembl Biomart, it just hourglasses for a while and then pops up a dialog box saying there appears to be a problem with my query. So it's not clear if this query is something you can do at this time, regardless.

I've had a more detailed look at this, and I don't think the issues are to do with selecting attributes from different pages or datasets.  To me it more seems that when the underlying ensembl database contains no phenotypic information associated with a SNP, Biomart simply doesn't return anything for that variant, no matter what other attributes you've asked for.  To demonstrate this, the code below looks for 3 SNPs, including the one in your original question.

library(biomaRt)
mart.snps <- useMart('ENSEMBL_MART_SNP', 'hsapiens_snp')

snps = c("rs368991480",
         "rs1082714",
         "rs1085093")

getBM(attributes = c("refsnp_id", 
                     "ensembl_gene_stable_id", 
                     "associated_gene"),
      filters = "snp_filter",
      values = snps,
      mart = mart.snps)

  refsnp_id ensembl_gene_stable_id associated_gene
1 rs1082714                                  CAND1
2 rs1085093        ENSG00000203943            GLTP

You can see that associated gene information is returned for the two new SNPs, and our original one is simply dropped.  This seems inconsistent since it doesn't happen to the intronic SNP (rs1082714) that doesn't have an associated ensembl gene id. To me this is obviously undesirable behaviour, but the issue is with ensembl / Biomart, rather than the biomaRt package itself.

Hi Mike,

Oh I see sorry, I got confused with an other associated gene name attribute. I guess what I gave above is a temporary solution then ;-).

Yes, this look like a bug from our side, rs368991480 seems to be missing from our BioMart phenotype table. I will investigate and see if we can get it fixed for our next release e!88.

Yes, you are right it's because rs1085093 is linked to two transcripts (We do our mapping from variants to Transcript). Maybe the package is doing some post-processing in the background, if you add the "ensembl_transcript_stable_id" attribute, you get the two transcripts back:

> getBM(attributes = c("refsnp_id", 
+                      "ensembl_gene_stable_id",
+                      "ensembl_transcript_stable_id",
+                      "associated_gene"),
+       filters = "snp_filter",
+       values = snps,
+       mart = mart.snps)
  refsnp_id ensembl_gene_stable_id ensembl_transcript_stable_id associated_gene
1 rs1082714                                                               CAND1
2 rs1085093        ENSG00000203943              ENST00000370673            GLTP
3 rs1085093        ENSG00000203943              ENST00000370671            GLTP