Please help me understand these values from the tutorial

I am having a hard time understanding the dG and ddG values that I obtain when going through the amino acid mutation tutorial.

  • I downloaded the dhdl files for the slow growth. Is this slow growth approach what is referred to as equilibrium thermodynamic integration (TI) in the literature? When I use gmx analyze -f forward_dhdl.xvg -integrate i obtain a value for dG=375964.46875 kJ/mol. This does not sound reasonable to me, what am i missing here?

  • I also downloaded the data for the fast growth and used -fA eqA/morphes/frame*/dgdl.xvg -fB eqB/morphes/frame*/dgdl.xvg -t 300 --nbins 25 to analyze it. I obtain a value for dG_folded=39.05 kJ/mol. The value for A2S mutation in the tripeptide from the website gives dG_unfolded=-39.15 kJ/mol. That would mean ddG=dG_folded-dG_unfolded=78.2kJ/mol. This is quite different from the provided experimental value of ddG=1.8kJ/mol. Am i misinterpreting this?

Any help to understand those discrepancies is very much appreciated. Thanks in advance!


gmx analyze just integrates. In this case, the x-axis is time (not lambda), which runs from 0 to 10000 (and not from 0 to 1 as lambda does). That way, you should divide the obtained value by 10000 to get a deltaG. This would be 37.6 kJ/mol in this case

I think you should look up S2A instead of A2S from the tripeptide database?



Dear Bert,

Thank you so much for your insightful response. It has been incredibly helpful.

You’re absolutely right; I should have used the S2A tripeptide instead of A2S. I apologize for that oversight. After making this correction, I obtain ddG=-0.1 ± 0.22 kJ/mol, which is still not within error for the experimental value of ddg=-1.8kJ/mol ( i expect its minus 1.8kJ/mol as it is stated to be mildly stabilizing mutation). But i guess this is as accurate as it gets? Or are there any more mistakes I made on the way? :slight_smile:


The computed ddG seems fine. Several things to keep in mind:

  • experimental error. Although in the tutorial (and in the paper) it is not explicitly provided, it should not be ignored that the experimental ddG has an associated uncertainty;
  • alchemical methods on average yield accuracy of ~1 kcal/mol, so the calculated ddG will not always match the experimental value exactly

thanks for the explanation!