| | Correlation (P value) |
|---|
| | 'Unbiased' | '2+' | '1cc' |
S
LR
| 0.54 (4.21 × 10-14) | 0.34 (1.50 × 10-11) | 0.03 (NS) |
S
BN
| 0.56 (< 2 × 10-16) | 0.33 (3.03 × 10-11) | 0.27 (0.0019) |
Pos | -0.26 (0.00299) | -0.23 (5.7 × 10-06) | -0.02 (NS) |
Foldedness | -0.24 (0.00564) | -0.19 (0.000137) | -0.25 (0.00555) |
Density | 0.43 (2.62 × 10-10) | 0.18 (0.00049) | 0.05 (NS) |
S
LR
+ pos | 0.52 (0.00818) | 0.37 (0.000552) | (NS) |
S
LR
+ pos + foldedness | 0.51 (0.0160) | 0.39 (0.00150) | (NS) |
- We calculated the Pearson correlation between the results of the kinase assay and either likelihood ratio score (S
LR
), the minimal product of binomial probabilities (S
BN
), the minimum distance from the either edge of the optimal cluster (identified using S
BN
) to the closest terminus (pos), the 'foldedness' of the optimal cluster, or simply the density of strong matches per residue (density). To calculate P values we used the generalized linear models implemented in R [57]. In addition, we fit linear models to combine the S
LR
score with the position and foldedness of the cluster (S
LR
+ pos and S
LR
+ pos + foldedness). When the variables did not all contribute significantly, we report NS (not significant). For the other sets, the P values are for the addition of the least significant term to the model. The total numbers of proteins in each set are slightly smaller than that reported [11] because since the time of that study proteins have been removed from the database and because scores cannot be computed for each gene for each method.
