Nature Podcast AI identifies gene interactions to speed up search for treatment targets
May 31, 2023
AI Snips
Chapters
Transcript
Episode notes
AI Predicts Gene Interactions
- Geneformer uses AI trained on massive cell transcriptomes to predict gene interactions.
- It identifies key genes influencing cell states, potentially revealing therapeutic targets.
Transfer Learning for Rare Diseases
- Transfer learning lets Geneformer apply knowledge from general data to rare diseases.
- It predicted candidate genes to revert diseased cells toward health in cardiomyopathy despite limited disease-specific data.
Lab Validates AI Predictions
- Lab tests confirmed Geneformer's predictions improved contracting ability of heart muscle cells.
- Additionally, the AI uncovered a previously unknown gene critical for cardiomyocyte contraction.