Preprint applies Alphagenome to map functional effects of CYP21A2 variants
A bioRxiv preprint uses reference-locus saturation analysis and the Alphagenome model to distinguish transcriptional, splicing, and protein-level consequences of single-nucleotide substitutions in CYP21A2, the gene responsible for most cases of congenital adrenal hyperplasia.
A preprint deposited on bioRxiv describes a systematic analysis of single-nucleotide variants in CYP21A2, which encodes steroid 21-hydroxylase and whose variants underlie the majority of congenital adrenal hyperplasia (CAH) cases. Interpreting CYP21A2 variants is complicated by the presence of a closely homologous pseudogene, complex allele structures, heterogeneous functional assay data, and uncertainty around noncoding effects — factors that frequently complicate clinical-grade variant classification in specialist centres.
The authors reconstructed the MANE select transcript (NM_000500.9) from GRCh38 genomic sequence and cross-referenced ClinVar, gnomAD, and UniProt variant annotations. They then applied Alphagenome, a recently described genomic deep-learning model, to perform a saturation-style in silico mutagenesis across the locus, predicting the transcriptional, splicing, and protein-structural consequences of each possible single-nucleotide substitution.
By integrating observed population variation with the model's predictions, the authors report the ability to assign candidate functional mechanisms to variants that currently lack confident classification. The preprint does not report experimental validation of the model's predictions. As this work has not yet undergone peer review, all findings should be regarded as preliminary.
Plain-language version
For patients, families, and general readers. Educational only — not medical advice.
Congenital adrenal hyperplasia (CAH) is a group of inherited conditions affecting the adrenal glands, which sit above the kidneys and make hormones the body needs to control salt, blood pressure, and stress responses. Most cases are caused by faults in a gene called CYP21A2. Identifying exactly which fault a person has can be difficult because the gene has a very similar-looking "decoy" nearby in the genome.
A new piece of research — not yet formally checked by other scientists — has used a computer model called Alphagenome to predict what different possible faults in CYP21A2 would do to the gene and the protein it makes. The aim is to help scientists and specialists understand which genetic changes matter and why. The work is at an early stage and has not yet been confirmed in laboratory experiments.
This is an educational summary, not medical advice. If anything here raises questions for you, please speak with your GP or a clinical professional.
Sources
Read the original reporting — these are the public sources this summary draws from.
-
Primary sourcePreprint bioRxiv (Cold Spring Harbor Laboratory) · 2026-10-16Analysis of single nucleotide substitution in CYP21A2 gene using Alphagenome