Lost in translation: the pitfalls of Ensembl gene annotations between human genome assemblies and their impact on diagnostics
Bibliographic record
Abstract
Abstract Background The GRCh37 human genome assembly is still widely used in genomics despite the fact an updated human genome assembly (GRCh38) has been available for many years. A particular issue with relevant ramifications for clinical genetics currently is the case of the GRCh37 Ensembl gene annotations which has been archived, and thus not updated, since 2013. These Ensembl GRCh37 gene annotations are just as ubiquitous as the former assembly and are the default gene models used and preferred by the majority of genomic projects internationally. In this study, we highlight the issue of genes with discrepant annotations, that have been recognized as protein coding in the new but not the old assembly. These genes are ignored by all genomic resources that still rely on the archived and outdated gene annotations. Moreover, the majority if not all of these discrepant genes (DGs) are automatically discarded and ignored by all variant prioritization tools that rely on the GRCh37 Ensembl gene annotations. Methods We performed bioinformatics analysis identifying Ensembl genes with discrepant annotations between the two most recent human genome assemblies, hg37, hg38, respectively. Clinical and phenotype gene curations have been obtained and compared for this gene set. Furthermore, matching RefSeq transcripts have also been collated and analyzed. Results We found hundreds of genes (N=267) that were reclassified as “protein-coding” in the new hg38 assembly. Notably, 169 of these genes also had a discrepant HGNC gene symbol between the two assemblies. Most genes had RefSeq matches (N=199/267) including all the genes with defined phenotypes in Ensembl genes GRCh38 assembly (N=10). However, many protein-coding genes remain missing from the current known RefSeq gene models (N=68) Conclusion We found many clinically relevant genes in this group of neglected genes and we anticipate that many more will be found relevant in the future. For these genes, the inaccurate label of “non-protein-coding” hinders the possibility of identifying any causal sequence variants that overlap them. In addition, Important additional annotations such as evolutionary constraint metrics are also not calculated for these genes for the same reason, further relegating them into oblivion.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.040 | 0.108 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.005 | 0.005 |
| Science and technology studies | 0.001 | 0.003 |
| Scholarly communication | 0.005 | 0.003 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.004 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".