Skip to main content
. 2016 Jun 17;32(19):2936–2946. doi: 10.1093/bioinformatics/btw361

Table 2.

Summary of protein-level and position-level cross validation on the S2648 dataset

Method Protein-levela
Position-levelb
γd σe γd σe
INPSc 0.52 1.26 0.54 1.28
mCSMc 0.51 1.26 0.54 1.23
STRUM 0.54 1.25 0.64 1.14

aProtein-level, mutation samples are divided based on their protein origin.

aPosition-level, mutation samples are divided based on their positions.

aINPS, mCSM, data taken from authors’ publications.

aγ, PCC between predicted and experimental ΔΔG.

aσ, RMSE of ΔΔG prediction in Kcal/mol.