8.6.1 Gene Identifier Systems

Since the first genetic sequence for a gene was determined in 1965 - an alanine tRNA in yeast(Holley et al. 1965) - scientists have been trying to give them names. In 1979, formal guidelines for the nomenclature used for human genes was proposed(Shows et al. 1979) along with the founding of the Human Gene Nomenclature Committee (Bruford et al. 2020), so that researchers would have a common vocabulary to refer to genes in a consistent way. In 1989, this committee was placed under the auspices of the Human Genome Organisation (HUGO), becoming the HUGO Gene Nomenclature Committee (HGNC). The HGNC has been the official body providing guidelines and gene naming authority since, and as such HGNC gene names are often called gene symbols.

Gene symbols are the most human-readable system for naming genes. The gene symbols BRCA1, APOE, and ACE2 may be familiar to you as they are involved in common diseases, namely breast cancer, Alzheimer’s Disease risk, and SARS-COV-2, respectively. Typically, the gene symbol is an acronym that roughly represents a label of what the gene is or does (or was originally found to be or do, as many genes are subsequently discovered to be involved in entirely separate processes as well), e.g. APOE represents the gene apolipoprotein E. This convention is convenient for humans when reading and identifying genes. However, standardized though the symbol conventions may be, they are not always easy for computers to work with, and ambiguities can cause mapping problems.

Other gene identifier systems developed along with other gene information databases. In 2000, the Ensembl genome browser was launched by the Wellcome Trust, a charitable foundation based in the United Kingdom, with the goal of providing automated annotation of the human genome. The Ensembl Project, which supports the genome browser, recognized even before the publication of the human genome that manual annotation and curation of genes is a slow and labor intensive process that would not provide researchers around the world timely access to information. The project therefore created a set of automated tools and pipelines to collect, process, and publish rapid and consistent annotation of genes in the human genome. Since its initial release, Ensembl now supports over 50,000 different genomes.

Ensembl assigns a automatic, systematic ID called the Ensembl Gene ID to every gene in its database. Human Ensembl Gene IDs take the form ENSG plus an 11 digit number, optionally followed by a period delimited version number. For example, at the time of writing the BRCA1 gene has an Ensembl Gene ID of ENSG00000012048.23. The stable ID portion (i.e. ENSGNNNNNNNNNNN) will remain associated with the gene forever (unless the gene annotation changes “dramatically” in which case it is retired). The .23 is the version number of the gene annotation, meaning this gene has been updated 22 times (plus its initial version) since addition to the database. The additional version information is very important for reproducibility of biological analysis, since conclusions drawn by results of these analyses are usually based on the most current information about a gene which are continually updated over time.

Ensembl Gene IDs and gene symbols are the most commonly used gene identifiers. The GENCODE project which provides consistent and stable genome annotation releases for human and mouse genomes, uses these two types of identifiers with its official annotations.

8.6.2 Mapping Between Identifier Systems

A very common operation in biological analysis is to map between gene identifier systems, e.g. you are given Ensembl Gene ID and want to map to the more human-recognizable gene symbols. Ensembl provides a service called BioMart that allows you to download annotation information for genes, transcripts, and other information maintained in their databases. It also provides limited access to external sources of information on their genes, including cross references to HGNC gene symbols and some other gene identifier systems. The Ensembl website has helpful documentation on how to use BioMart to download annotation data using the web interface.

In addition to downloading annotations as a CSV file from the web interface, Ensembl also provides the biomaRt Bioconductor package to allow programmatic access to the same information directly from R. There is much more information in the Ensembl databases than gene annotation data that can be accessed via BioMart, but we will provide a brief example of how to extract gene information here:

# this assumes biomaRt is already installed through bioconductor
library(biomaRt)

# the human biomaRt database is named "hsapiens_gene_ensembl"
ensembl <- useEnsembl(biomart="ensembl", dataset="hsapiens_gene_ensembl")

# listAttributes() returns a data frame, turn into a tibble to help with formatting
as_tibble(listAttributes(ensembl))
# A tibble: 3,143 x 3
   name                          description                  page        
   <chr>                         <chr>                        <chr>       
 1 ensembl_gene_id               Gene stable ID               feature_page
 2 ensembl_gene_id_version       Gene stable ID version       feature_page
 3 ensembl_transcript_id         Transcript stable ID         feature_page
 4 ensembl_transcript_id_version Transcript stable ID version feature_page
 5 ensembl_peptide_id            Protein stable ID            feature_page
 6 ensembl_peptide_id_version    Protein stable ID version    feature_page
 7 ensembl_exon_id               Exon stable ID               feature_page
 8 description                   Gene description             feature_page
 9 chromosome_name               Chromosome/scaffold name     feature_page
10 start_position                Gene start (bp)              feature_page
# ... with 3,133 more rows

The name column provides the programmatic name associated with the attribute that can be used to retrieve that annotation information using the getBM() function:

# create a tibble with ensembl gene ID, HGNC gene symbol, and gene description
gene_map <- as_tibble(
  getBM(
    attributes=c("ensembl_gene_id", "hgnc_gene_symbol", "description"),
    mart=ensembl
  )
)
gene_map
# A tibble: 68,012 x 3
   ensembl_gene_id hgnc_symbol description                                                               
   <chr>           <chr>       <chr>                                                                     
 1 ENSG00000210049 MT-TF       mitochondrially encoded tRNA-Phe (UUU/C) [Source:HGNC Symbol;Acc:HGNC:748~
 2 ENSG00000211459 MT-RNR1     mitochondrially encoded 12S rRNA [Source:HGNC Symbol;Acc:HGNC:7470]       
 3 ENSG00000210077 MT-TV       mitochondrially encoded tRNA-Val (GUN) [Source:HGNC Symbol;Acc:HGNC:7500]
 4 ENSG00000210082 MT-RNR2     mitochondrially encoded 16S rRNA [Source:HGNC Symbol;Acc:HGNC:7471]       
 5 ENSG00000209082 MT-TL1      mitochondrially encoded tRNA-Leu (UUA/G) 1 [Source:HGNC Symbol;Acc:HGNC:7~
 6 ENSG00000198888 MT-ND1      mitochondrially encoded NADH:ubiquinone oxidoreductase core subunit 1 [So~
 7 ENSG00000210100 MT-TI       mitochondrially encoded tRNA-Ile (AUU/C) [Source:HGNC Symbol;Acc:HGNC:748~
 8 ENSG00000210107 MT-TQ       mitochondrially encoded tRNA-Gln (CAA/G) [Source:HGNC Symbol;Acc:HGNC:749~
 9 ENSG00000210112 MT-TM       mitochondrially encoded tRNA-Met (AUA/G) [Source:HGNC Symbol;Acc:HGNC:749~
10 ENSG00000198763 MT-ND2      mitochondrially encoded NADH:ubiquinone oxidoreductase core subunit 2 [So~
# ... with 68,002 more rows

With this Ensembl Gene ID to HGNC symbol mapping in hand, you can combine the tibble above with other tibbles containing gene information by joining them together.

8.6.3 Mapping Homologs

Sometimes it is necessary to link datasets from different organisms together by orthology. For example, in an experiment performed in mice, we might be interested in comparing gene expression patterns observed in the mouse samples to a publicly available human dataset. In these contexts, we must link gene identifiers from one organism to their corresponding homologs in the other. BioMart enables us to extract these linked identifiers by explicitly connecting different biomaRt databases with the getLDS() function:

human_db <- useEnsembl("ensembl", dataset = "hsapiens_gene_ensembl")
mouse_db <- useEnsembl("ensembl", dataset = "mmusculus_gene_ensembl")
orth_map <- as_tibble(
  getLDS(attributes = c("ensembl_gene_id", "hgnc_symbol"),
       mart = human_db,
       attributesL = c("ensembl_gene_id", "mgi_symbol"),
       martL = mouse_db
  )
)
orth_map
# A tibble: 26,390 x 4
   Gene.stable.ID  HGNC.symbol Gene.stable.ID.1   MGI.symbol
   <chr>           <chr>       <chr>              <chr>     
 1 ENSG00000198695 MT-ND6      ENSMUSG00000064368 "mt-Nd6"  
 2 ENSG00000212907 MT-ND4L     ENSMUSG00000065947 "mt-Nd4l"
 3 ENSG00000279169 PRAMEF13    ENSMUSG00000094303 ""        
 4 ENSG00000279169 PRAMEF13    ENSMUSG00000094722 ""        
 5 ENSG00000279169 PRAMEF13    ENSMUSG00000095666 ""        
 6 ENSG00000279169 PRAMEF13    ENSMUSG00000094741 ""        
 7 ENSG00000279169 PRAMEF13    ENSMUSG00000094836 ""        
 8 ENSG00000279169 PRAMEF13    ENSMUSG00000074720 ""        
 9 ENSG00000279169 PRAMEF13    ENSMUSG00000096236 ""        
10 ENSG00000198763 MT-ND2      ENSMUSG00000064345 "mt-Nd2"  
# ... with 26,380 more rows

The mgi_symbol field refers to the gene symbol assigned in the Mouse Genome Informatics database maintained by The Jackson Laboratory.