Table 2 Splicing-related pathogenicity predictor comparison
Splicing specific | Model type | Features | Prediction target | Constraints | |
|---|---|---|---|---|---|
CADD | No | SVM & Logistic Regression | 1029 | True vs Simulated Variants | – |
IntSplice2 | Yes | LightGBM | ~100 | Pathogenicity (ClinVar) | Intronic SNVs (3–50nts inside intron) |
MMSplice | Yes | Neural Nets (MLP, CNNs) & Logistic Regression | Sequence Context | Splicing Effects & Pathogenicity (ClinVar) | 6 region specific models with unique architectures |
TraP | Yes | SVM | 56 | Pathogenicity (ClinVar) | SNVs |
RegSNPs-Intron | Yes | Random Forest | 436 | Pathogenicity (ClinVar) | Intronic SNVs (3–50nts inside intron) |
RegSNPs-Splicing | Yes | Random Forest | ~450 | Pathogenicity (ClinVar) | Synonymous SNVs |
S-CAP | Yes | Gradient Boosting Tree | ~31 | Pathogenicity (ClinVar) | SNVs |
