Table 2.
The summary of prediction scores for nonsynonymous substitutions in the GCK amino acid sequence, for which data were available either on their clinical phenotype, the GSIR-T or the Hill coefficient. For a detailed list of nonsynonymous substitutions analyzed, relevant raw data and references, cf. Table S2.
| Measure | Prediction method: Variable | SIFT | PolyPhen2 | PhD-SNP | PoPMuSiC 2.1 | SNAP2 | SNPs&GO (GO terms excluded) | SNPs&GO (GO terms included) | I-Mutant 3 | Align GVGD | EVmutation |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Disease | |||||||||||
| PHHI (n = 16) | Sensitivity [%] | 63 | 94 | 44 | 94 | 25 | 13 | 56 | 81 | 81 | 75 |
| Non-diabetic (n = 16) | Sensitivity [%] | 75 | 38 | 75 | 27 | 88 | 94 | 50 | 20 | 50 | N/A (53) |
| Non-diabetic (n = 515) | False positive ratio [%] | 13.2 | 5.2 | 15.7 | 5.5 | 23.7 | 28.0 | 11.1 | 15.5 | 9.5 | 3.6 |
| MODY (n = 499) | Sensitivity [%] | 87.6 | 94.8 | 85.6 | 94.5 | 78.0 | 66.9 | 87.0 | 84.6 | 75.6 | 97.1 |
| GSIR-T | |||||||||||
| <4 (n = 19) | Sensitivity [%] | 63 | 95 | 53 | 100 | 32 | 13 | 39 | 84 | 84 | 72 |
| 4–5.5 (n = 20; normal range) | Sensitivity [%] | 25 | 10 | 35 | 15 | 40 | 55 | 30 | 20 | 15 | 5 |
| 4–5.5 (n = 106) | False positive ratio [%] | 28 | 3 | 15 | 4 | 27 | 29 | 16 | 14 | 7 | 8 |
| >5.5 (n = 87) | Sensitivity [%] | 92 | 98 | 93 | 95 | 82 | 71 | 89 | 86 | 95 | 97 |
| Hill coefficient n H | |||||||||||
| <1.2 (n = 23) | Sensitivity [%] | 96 | 91 | 96 | 100 | 87 | 74 | 96 | 87 | 96 | 100 |
| 1.2–1.5 (n = 48) | Sensitivity [%] | 83 | 96 | 79 | 94 | 69 | 50 | 77 | 88 | 94 | 94 |
| >1.5 (n = 56; normal range) | Sensitivity [%] | 18 | 4 | 23 | 5 | 36 | 39 | 23 | 14 | 11 | 11 |
| >1.5 (n = 71) | False positive ratio [%] | 13 | 4 | 15 | 4 | 25 | 30 | 15 | 13 | 6 | 4 |
PoPMuSiC2.1 and I-Mutant 3 were only calculated for amino acids available in the crystal structure (PDB ID: 1V4S). Calculations for some amino acids were not available for the EVmutation. The threshold value for EVmutation was set to a median value of normoglycemic variations (−2.39) as all but two neutral nonsynonymous substitutions exceeded the originally suggested zero threshold11.