Article ID: 1090 - Last Modified: October 2, 2013
I have a receptor/ligand pair with an unknown experimental binding energy, and I want to use Liaison to determine the computed binding energy. How do I calculate the ligand's binding energy, and how do I ensure that the calculated value for the ligand/receptor pair is realistic?
In order to use Liaison to determine the unknown binding energy of a ligand with a specific receptor, you must first have a training set of known binding energies of a set of ligands (the more the better) with that receptor. If you do not have a training set, you can use Glide's "score-in-place" option to get a rough estimate of the binding affinity.
Once you have collected a training set, run a Liaison job with the appropriate sampling method and job conditions. The Liaison output includes calculated values for the van der Waals, electrostatic, and cavity energy terms for the free and bound state of the ligands as seen below:
Delta G = alpha * (U vdw_b - U vdw_f) + beta * (U elec_b - U elec_f) + gamma * (U cav_b - U cav_f)
- Delta G = calculated binding energy
- U xxx_b = the van der Waals, Coulombic and Cavity energy terms from the bound state Liaison calculation (bound.inp)
- U xxx_f = the van der Waals, Coulombic and Cavity energy terms from the free state Liaison calculation (free.inp)
- alpha, beta, gamma = the coefficients of their respective energy terms
Using the above Liaison-determined energy terms, run Strike to calculate the values of alpha, beta, and gamma using the known ligand/receptor binding energies.
The average GlideScore is computed in Liaison simulations along with the van der Waals, Coulombic, and Cavity energy terms. The linear fitting of the computed GlideScores to the experimental binding energies involves two terms, the slope and intercept.
At this point, determine the quality of the fitting results by directly examining (plotting, for example) the calculated and empirical binding energy pairs or by examining the metrics produced by Strike, such as the R-squared or the RMSD values. Once you feel that the values of the fitting parameters are within acceptable limits, use them with the Liaison simulation results for the unknown ligand (which is generally run using the same simulation conditions as the training ligand set). The Strike prediction task then produces a calculated binding energy value for the receptor and the unknown ligand.
Determining whether or not the obtained binding energy value is realistic requires chemical intuition, experience, and knowledge of the system in question. Liaison does not attempt to determine the accuracy of binding energy values; it simply presents a set of answers given variations in job conditions.
#573: Can I calculate binding affinities with Schrödinger software?
Type the words or phrases on which you would like to search, or click here to view a list of all
Knowledge Base articles