LRG project

Created specifically for clinical reporting by manual curation

In recognition of the need to create universally accepted reference standards for variant reporting, a meeting of key stakeholders, including EMBL-EBI, the NCBI, HGVS, LSDB curators and other members of the community, was held in 2008. The goal of the meeting was to design a reference system that would address the shortcomings of existing systems, including confusion over versioning, and that would allow consistent and unambiguous reporting of variants in clinically relevant loci. The new system, founded on the RefSeqGene project, was named Locus Reference Genomic (LRG).

As of August 2019, over 1323 LRGs have been created, of which over 859 are public and in use by the community.
The aim of the project is to create an LRG for every locus with clinical implications.


Accurate and unambiguous reporting of variants over time is difficult, partly due to regular updates to transcripts and the genome reference assembly. Confusion over different transcript versions has also resulted in inconsistent variant reporting in the past. To address these challenges, all reference sequences included in the fixed section of public LRG records are stable, thus they neither change over time nor version. As a result, the LRG project establishes a coordinate system that is independent from upgrades to the genome reference assembly and includes sequences that are independent of ongoing changes to transcripts. The LRG system also establishes identifiers that do not version or change. To facilitate reporting, each LRG record provides annotation and mapping data that define the relationship between the LRG fixed section sequences, the genome reference assembly (GRCh37 and GRCh38) and all known transcripts from RefSeq and GENCODE.

Flexible and collaborative

Each LRG record is created in collaboration with experts in the region of interest. Therefore, inclusion of legacy or community-requested reference sequences is supported. However, in recognition of the fact that NGS technology is based on alignments to the genome reference assembly, it is now our policy for all LRGs to match the genome reference assembly (GRCh38), unless there is a known error in the genome reference assembly or there is a prevalent legacy alternate allele that has been consistently and systematically used in clinical reporting.

Connect the past, present and future of clinical variant reporting

NCBI (RefSeq) and EMBL-EBI (Ensembl/GENCODE) are working together to rationalise differences in our gene sets. Our goals for 2018/2019 include faster convergence on key high value annotations to provide a common minimal set of transcripts per gene. The goal for this transcript subset is identity between RefSeq and Ensembl both in terms of transcript length and per each nucleotide base. We encourage the community to continue working with us to create LRGs, which will have input into this future goal.

To be compatible with NGS-based sequencing, we aim to match all LRGs to the genome reference assembly (GRCh38). However, when an LRG based on legacy sequence is required, creating the LRG defines the relationship of the legacy sequence to the reference genome assembly. This facilitates the usage of legacy data alongside NGS technology. Where the community requests an allele that is incompatible with the genome reference assembly, we evaluate and request a review by the GRC if appropriate.

The annotation for both RefSeq and GENCODE is reviewed as part of the manual curation process. Once a transcript model has been selected, efforts are made to have the corresponding RefSeq and GENCODE transcripts match each other and match the reference. In addition, we aim to have complete identity between the two transcripts, including the UTRs. This facilitates unambiguous multi-directional data exchange between RefSeq, GENCODE and the reference genome assemblies (GRCh37, GRCh38).

LRG records contain both GENCODE and RefSeq transcripts in the updatable section, therefore they are a unique record of annotation from both transcript sets in one place. More information on sequence sources can be found here.


All LRGs are generated and maintained by the NCBI and EMBL-EBI. They can be visualized in the Ensembl, NCBI and UCSC genome browsers in context with all other existing annotations. Also, LRG sequences are compatible with HGVS nomenclature are supported by nomenclature checker systems, such as Mutalyzer and VariantValidator. Use of LRG records has been recommended by key organisations, such as the Human Genome Variation Society (HGVS), the European Molecular Genetics Quality Network (EMQN), and the American College of Medical Genetics and Genomics (ACMG).