Annotating Non-Coding Loci in Ensembl and Ensembl Genomes

Ensembl provides reference gene annotation on vertebrate genomes while Ensembl Genomes both creates and displays gene annotation for bacterial, fungal, metazoan and plant species. While only a relatively few functional ncRNAs have been extensively characterised, there remains an enormous interest in the identification and classification of novel long- and small non-coding RNAs. Long non-coding RNAs in human and mouse are annotated by both manual and automated methods under the umbrella of the GENCODE consortium. Manual annotators assess alignments of a wide range of transcriptomic data types including RNAseq, PacBio and SLRSeq to identify support for lncRNA transcripts while the Ensembl gene build pipeline annotates lncRNAs by first generating transcript models from transcriptome data, followed by filtering to remove models with protein coding potential. For human and mouse, these models are manually validated by the HAVANA team, but this additional step is not available for most vertebrate genomes. For Ensembl Genomes, lncRNA annotation are displayed if they are provided by a data submitter but no further annotation is added where this is not the case. In addition to cataloguing lncRNA models in existing datasets, GENCODE are improving the annotation of lncRNAs in human and mouse using capture-seq technology and PacBio sequencing to extend existing annotation and identify novel full-length transcripts and loci. GENCODE also investigate the coding potential of all lncRNA loci using proteomics data and conservation of coding potential highlighted by PhyloCSF.

Small RNAs are annotated exclusively in Ensembl using automated pipelines, although where available, additional information is incorporated. For human and mouse miRBase annotations of miRNAs are directly imported, but for other species miRNA genes are annotated by an initial BLASTN search of genomic sequence against miRBase stem-loop sequences followed by filtering. Gene models are then created when RNAFold infers that the underlying genomic sequence forms a stable hairpin structure. Additional small ncRNA genes are annotated by first aligning sequences from Rfam to the genome using BLASTN followed by processing using the Infernal suite of programs. Where small ncRNA gene annotations are present in gene sets provided to Ensembl Genomes they are displayed. However, in general small ncRNA genes are annotated by Ensembl Genomes in a similar way to the Ensembl pipeline. Alignments of the latest Rfam models against all plant and metazoan genomes are used to eliminate the need for the BLASTN filtering and taxonomic filters reduce inappropriate false positive hits.