This is an important question! A good reference can be found at
https://pmc.ncbi.nlm.nih.gov/articles/PMC9662604/
Figure 13 makes the relevant point -- You should always plot dG vs
-log(x) (where x is Kd, Ki, or IC50). It is also reasonable to shift the
computed dGs such that their mean matches the reference. This has no
effect on the correlation values but minimizes the RMSE.
If you plot ddG you can get randomly different correlation statistics
depending on which graph edges you (semi arbitrarily) chose.
HTH,
BKR
On 1/13/25 5:52 PM, King Wang via AMBER wrote:
> Hello AMBER fellows,
>
> I would like to ask or have opinions on how people usually report their ΔΔG
> relative binding free energy vs experimental ΔG when they investigate free
> energy calculation using FEP+ or TI calculations using AMBER.
>
> I have included some figures in the articles below, as people report the ΔG
> (binding free energy for each ligand) in the Y-axis vs experimental ΔG in
> the x-axis using TI, https://dx.doi.org/10.1021/acsomega.9b04233
> [image: Screenshot 2025-01-13 at 1.26.30 PM.png]
> or below, DOI: 10.1021/acs.jcim.9b00105, using TI and FEP+
> [image: Screenshot 2025-01-13 at 1.29.15 PM.png]
> My understanding for reported experimental ΔGs were from the IC50, Kd, Ki,
> and convert them to ΔG using -RTln(Kd), and this should be the binding
> affinity for the corresponding ligand, whereas when we simulate/calculate
> the relative binding free energy from ligand 1 to ligand2, there is not
> absolute, but relative binding free energy ΔΔG?! Then, how should we report
> in terms of graphing? ΔΔG in Y axis and experimental ΔG in X axis? I am not
> sure about that?
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Jan 14 2025 - 11:30:02 PST