CADD-Splice—improving genome-wide variant effect prediction using deep learning-derived splice scores
importance 7/10
partially solved
Genomics and Rare Diseases genomic diagnostics stage: research
Problem
Genetic variants affecting splicing are hard to identify beyond donor/acceptor dinucleotides, underpinning much genetic disease.
Why it matters
Missed splice-disrupting variants delay diagnosis and treatment for patients with genetic disease.
What the AI did
Deep neural networks were used to derive splice scores, compared multiple ML methods, and integrated the best into the CADD genome-wide variant effect predictor.
Results
Integrated two specialized splicing scores into CADD; abstract truncated before full metrics.
Record
| Venue | Genome Medicine |
|---|---|
| Published | 2021 |
| Work type | article |
| Authors | Philipp Rentzsch, Max Schubach, Jay Shendure, Martin Kircher |
| Citations | 637 (OpenAlex, at import time) |
| References | 64 |
| Model or method | Deep neural networks; CADD |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1186/s13073-021-00835-9 |
| Open access | gold |
| Keywords | splicing, deep learning, variant effect prediction, CADD, genomics |
Importance rating
7
significant — thousands to millions affected
Abstract states splicing-impacting variants underlie a substantial proportion of genetic disease; missed variants affect diagnosis across all genetic disease patients.