Skip to main content
. 2023 Dec 22;12:e84756. doi: 10.7554/eLife.84756

Table 1. Correlation between Flex ddG predictions for 250 runs and experimental data (see table 4 of Sirawaraporn et al., 1997) for P. falciparum dihydrofolate reductase (PfDHFR) pyrimethamine resistance mutations.

Mutation ΔΔGexp*(kcal/mol) Exp. sum Exp I.E. ΔΔGFlexddG §(kcal/mol) Sum I.E.§**
N51I –0.783 –0.124
C59R –0.184 –0.033
S108N 1.297 0.312
I164L –0.351 –0.323
N51I,S108N 1.89 0.514 1.376 –0.166 0.188 –0.354
C59R,S108N 2.29 1.113 1.177 0.399 0.279 0.119
N51I,C59R,S108N 2.595 0.33 2.265 0.162 0.155 0.007
C59R,S108N,I164L 3.283 0.762 2.521 0.018 –0.043 0.061
N51I,C59R,S108N,I164L 3.761 –0.021 3.782 0.301 –0.168 0.469
Pearson correlation 0.611 0.660 0.756
Correctly classified 8/9 4/5 4/5
*

Experimentally measured PfDHFR pyrimethamine binding free energy change data from Sirawaraporn et al., 1997.

Sum of experimental values of binding free energy change for independent mutations.

Interaction energy calculated as the difference between experimentally measured values of binding free energy change of multiple mutant compared to the sum of the independent mutations involved.

§

Change in PfDHFR pyrimethamine binding free energy predicted by Flex ddG calculated as the average of the distribution of runs. Free energy predictions from Rosetta are in Rosetta Energy Units, however the authors of Flex ddG applied a generalised additive model to reweight the predictions and make the output more comparable to units of kcal/mol (Barlow et al., 2018).

Sum of Flex ddG predictions for independent mutations.

**

Interaction energy calculated as the difference between Flex ddG predicted binding free energy change of multiple mutant compared to the sum of the independent mutations.