Skip to main content

Table 1 Performances in 5-fold validation

From: PoPMuSiC 2.1: a web server for the estimation of protein stability changes upon mutation and sequence optimality

 

All mutants

Exclusion of 10% outliers

 

R d / R c a

σ d /σ c (kcal/mol) b

R d / R c a

σ d /σ c (kcal/mol) b

ΔΔGP = < ΔΔGM >

-

1.47

-

1.10

Random predictorc

0.00 (0.02)

2.08 (0.02)

0.22 (0.02)

1.60 (0.02)

Run 1

0.64/0.62

1.13/1.16

0.79/0.78

0.86/0.89

Run 2

0.63/0.64

1.15/1.08

0.79/0.78

0.87/0.85

Run 3

0.65/0.59

1.11/1.20

0.80/0.78

0.85/0.88

Run 4

0.64/0.61

1.12/1.18

0.80/0.77

0.85/0.88

Run 5

0.64/0.63

1.13/1.15

0.79/0.78

0.85/0.89

Average 1 d

0.64/0.62

1.13/1.16

0.79/0.78

0.85/0.88

Average 2 d, e

-/0.63

-/1.15

-/0.79

-/0.86

  1. a Correlation coefficient between predicted and measured ΔΔG's, in the training (Rd) and validation (Rc) set. b root mean square error between predicted and measured ΔΔG's, in the training (Rd) and validation (Rc) set. e The random predictor is obtained by using a randomly shuffled set of ΔΔGM values as predicted ΔΔG values. The average (standard deviation) values of R and σ over 1000 runs are given. d Average values of R and σ, over the five different runs. e Badly modeled mutants are removed from the training sets before parameter identification, but they are maintained in the validation sets.
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us