Skip to contents

Gene Identifier Handling

2026-08-24

Source: vignettes/gene_identifiers.Rmd
gene_identifiers.Rmd

RNAsum uses Ensembl gene IDs as its internal identifier throughout processing, then converts to HGNC gene symbols for display in the final report. The reference genome is GRCh38 with Ensembl annotation version 105.

Input identifiers

Patient expression data (Salmon or Kallisto) must use Ensembl gene or transcript IDs:

  • quant.genes.sf (Salmon gene-level): Ensembl gene IDs are read directly from the Name column. Version suffixes (e.g. ENSG00000012048.21) are stripped automatically.
  • quant.sf (Salmon transcript-level) and abundance.tsv (Kallisto): Ensembl transcript IDs are collapsed to gene-level Ensembl IDs via tximport, using a pre-built transcript-to-gene mapping derived from Ensembl v105 via AnnotationHub (stored in the companion RNAsum.data package).

PAR_Y pseudo-autosomal loci (genes duplicated on both X and Y chromosomes) are filtered out at input to prevent row-name conflicts when merging with reference data.

The TCGA reference cohorts bundled with RNAsum are also keyed on Ensembl gene IDs, so sample and reference data join on a shared, stable identifier space.

Internal processing

All count matrices (sample + reference) are retained with Ensembl IDs through normalisation, filtering, batch correction, and PCA.

Annotation and symbol conversion

After processing, genes are annotated using the same Ensembl v105 mapping table (tx_gene_id_105.rds), which provides:

Field Description
ENSEMBL Ensembl gene ID (e.g. ENSG00000012048)
SYMBOL HGNC gene symbol (e.g. BRCA1)
GENEBIOTYPE Ensembl biotype (e.g. protein_coding, lncRNA)
SEQNAME Chromosome
GENESEQSTART / GENESEQEND Genomic coordinates

Two deduplication steps are applied:

  1. Duplicate Ensembl IDs — removed (keeps first occurrence).
  2. Duplicate gene symbols — removed (affects multi-locus genes such as Y_RNAs, SNORDs, and LINC genes that map to multiple genomic locations).

Genes with no HGNC symbol in the Ensembl v105 annotation are silently excluded from the reportable set.

After annotation, the processed expression matrix is re-indexed on HGNC gene symbols — this is the identifier used in all HTML report sections (Mutated genes, Fusion genes, CN altered genes, Cancer genes, etc.) and all output tables.

Summary 

Ensembl transcript/gene IDs (input)
        ↓  tximport (transcript-level only)
Ensembl gene IDs  ──────────────────────────── internal processing
        ↓  Ensembl v105 annotation (tx_gene_id_105.rds)
HGNC gene symbols (report display and output tables)

The annotation reference (tx_gene_id_105.rds) is versioned and fixed — it does not update automatically. GRCh37 is not supported.