Machine Learning approaches for the detection of disease-causing variants in whole-genome data need to address the expression of functional genes
Camilla Mapstone, Julia Handl, David Talavera · 2026
Read Machine Learning approaches for the detection of disease-causing variants in whole-genome data need to address the expression of functional genes by Camilla Mapstone, Julia Handl, David Talavera (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0355557.txt.
Description
Camilla Mapstone, Julia Handl, and David Talavera stress-test the hope that machine learning on whole genomes can find disease-causing gene combinations. Dosage effects across many genes are thought to drive disease, but the search space is enormous. The team built benchmarks two ways. Real data came with known answers. Synthetic data came with dials for size and noise. Standard models were then trained to classify patients and to name the guilty genes.
A plain regression model proved most reliable at fingering the true genes across synthetic worlds, even when its raw classification score lagged flashier rivals. Attempts on a major biobank dataset failed outright, and digging in showed why: models stumbled unless the activity of working genes was taken into account. Mapstone and colleagues say genome-wide learners must respect which genes are actually switched on.
Catalog record
| CREATOR | Camilla Mapstone, Julia Handl, David Talavera |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 59878 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |