Table 1.

Accuracy of using mutation frequency or change in structural stability to predict viral infectivity in binary classification manner.

Predictor	Sensitivity a	Specificity b	Precision c	Accuracy d
Mutation Frequency e	59.5%	70.0%	67.10%	64.10%
Stability of Reference Models f	73.33%	78.72%	76.74%	76.12%
Composite Score g	80.00%	79.57%	78.16%	79.78%

^a Sensitivity = (True positive)/(True positive + False negative); ^b Specificity = (True negative)/(True negative + False positive); ^c Precision = (True positive)/(True positive + False positive); ^d Accuracy = (True positive + True negative)/(True positive + True negative + False positive + False negative); ^e Mutations with a database frequency of 0.2% or less are predicted to result in non-infectious virus; ^f Mutants with structural stability higher than the reference models are predicted to result in non-infectious virus; ^g Mutants with a composite score (sum of ranks of frequency and FOLDEF stability) higher than 175 are predicted to result in non-infectious virus.