Article ID: 763 - Last Modified: June 12, 2011
Can the Glide settings be changed so as to obtain a correlation between the score values and their experimental Ki and IC50 values?
Glide has been designed for docking accuracy and database enrichment (i.e, distinguishing actives from inactives), not ranking actives. It is difficult to produce strong correlations with experimental binding energies with Glide because of the approximations introduced, such as the rigid receptor, the use of grids for the potential. These approximations make it possible to scren millions of compounds with Glide and obtain enrichment of actives.
Computing absolute binding energies of actives, or even predicting the relative ranking, is an extremely challenging task. If you want to compute binding energies and obtain rank ordering, you could consider using Prime MM-GBSA, which has been effective in rank-ordering a congeneric series of compounds; Liaison, which uses experimental data for calibration of a response model; and MacroModel Embrace, which allows you to do a conformational search of the bound and the free ligand to obtain binding energies.
If you have a set of known actives, you can develop your own scoring function by fitting a model to the experimental data with Strike (for example). The data could come from Glide, or by running the Ligand and Structure-Based Descriptors workflow, for example.
Keywords: GlideScore, experimental Ki values, IC50 value, custom scoring function, Maestro
#793: How accurate are the XP GlideScores?
#573: Can I calculate binding affinities with Schrödinger software?
#936: It is possible to calculate the IC50 value of a ligand-receptor complex with any of the Schrödinger software?
#144: GlideScore/Docking Score doesn't correlate with my known activities. What is wrong?
#1027: What is the difference between GlideScore and Emodel, and which should I use for ranking poses?
Type the words or phrases on which you would like to search, or click here to view a list of all
Knowledge Base articles