Real-world evaluation of deep learning algorithms to classify functional pathogenic germline variants
Chow, R. D.; Parikh, R. B.; Nathanson, K. L.
Show abstract
Deep learning models for variant pathogenicity prediction can recapitulate expert-curated annotations, but their performance remains unexplored on actual disease phenotypes in a real-world setting. Here, we apply three state-of-the-art pathogenicity prediction models to classify hereditary breast cancer gene variants in the UK Biobank. Predicted pathogenic variants in BRCA1, BRCA2 and PALB2, but not ATM and CHEK2, were associated with increased breast cancer risk. We explored gene-specific score thresholds for variant pathogenicity, finding that they could improve model performance. However, when specifically tasked with classifying variants of uncertain significance, the deep learning models were generally of limited clinical utility.
Matching journals
The top 4 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Performance of polygenic risk scores for cancer prediction in a racially diverse academic biobank 94%
- Classification of Variants of Reduced Penetrance in High Penetrance Cancer Susceptibility Genes: Framework for Genetics Clinicians and Clinical Scientists by CanVIG-UK (Cancer Variant Interpretation Group-UK) 94%
- Impact of genetic counselling strategy on diagnostic yield and workload for whole genome sequencing-based tumour diagnostics 92%
Similar papers in this journal
- Segregation analysis of 17,425 population-based breast cancer families: evidence for genetic susceptibility and risk prediction 95%
- The contribution of coding variants to the heritability of multiple cancer types using UK Biobank whole-exome sequencing data 95%
- Availability of benign missense variant “truthsets” for validation of functional assays: current status and a novel systematic approach 95%
Similar papers in this journal
- Characteristics predicting reduced penetrance variants in the high-risk cancer predisposition gene TP53 92%
- Rare coding variants in five DNA damage repair genes associate with timing of natural menopause 92%
- Pleiotropy-guided transcriptome imputation from normal and tumor tissues identifies new candidate susceptibility genes for breast and ovarian cancer 92%
Similar papers in this journal
- Identifying therapeutic targets for cancer: 2,094 circulating proteins and risk of nine cancers 94%
- Whole-genome analysis of Nigerian patients with breast cancer reveals ethnic-driven somatic evolution and distinct genomic subtypes 94%
- Pan-cancer analysis demonstrates that integrating polygenic risk scores with modifiable risk factors improves risk prediction 94%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.