Work

Assessment of ability of a DNA language model to predict pathogenicity of rare coding variants

Year: 2025

Type: article

Abstract: A recently described method to predict pathogenicity of DNA variants uses a DNA language model and can be applied to both coding and non-coding variants. For coding variants the performance of this method, termed GPN-MSA (genomic pretrained network with multiple-sequence alignment), was reported to be superior to CADD. We compare the performance of this method against 45 other predictors applied to rare coding variants in 18 gene-phenotype pairs. We find that while GPN-MSA produces stronger evidence for association than CADD it is not the best-performing method for any gene and on average other prediction methods are superior. While GPN-MSA may be useful for predicting the pathogenicity of non-coding variants, it would seem sensible for clinicians and researchers to utilise other methods when dealing with coding variants.This research has been conducted using the UK Biobank Resource.

Source: Journal of Human Genetics

DOI: 10.1038/s10038-025-01385-3

Author: David Curtis

Institution: University College London

Language: English

Topic: Genomics and Rare Diseases

Subfield: Genetics

Field: Biochemistry, Genetics and Molecular Biology

Domain: Life Sciences

SDG: Good health and well-being

Open Access status: hybrid

Estimated APC paid: $5,490