Skip to main content
Fig. 4 | Genome Medicine

Fig. 4

From: Genome-wide prediction of pathogenic gain- and loss-of-function variants from ensemble learning of a diverse feature set

Fig. 4

Explanation of LoGoFunc predictions. a SHAP values by class for features with combined SHAP values in the 90th percentile and above. b (Top) The SHAP values for the top ten features for the seven GOF variants found in the ion channel SCN2A in the test set. (Middle) The SHAP values for the top ten features for the eight LOF SCN2A variants in the test set. (Bottom) The SHAP values for the top ten features for the seven neutral SCN2A variants in the test set. c The experimentally determined structure of SCN2A [64] with the represented GOF (red), LOF (blue), and neutral (yellow) SCN2A variants from the test set. d The SCN2A model from the AlphaFold2 prediction database annotated with the represented GOF (red), LOF (blue), and neutral (yellow) SCN2A variants from the test set

Back to article page

Genome Medicine

ISSN: 1756-994X

Contact us