Enlarge

Above, the SAH/MTA nucleoside receptor above the LSBD ligand set. A single LSBD run was used to generate multiple descriptor types for the entire ligand set, which is shown here with carbon atoms colored according to the compound‘s binding energy.

Ligand & Structure-Based Descriptors: A practical, efficient solution for rank-ordering compounds and predicting binding affinities

Rank-ordered predictions of ligand binding affinities can be invaluable for drug discovery, but creating a reliably predictive model can be a time-consuming endeavor with no guarantee of success. Schrödinger's Ligand & Structure-Based Descriptors (LSBD) module streamlines this process by integrating multiple computational applications under a single interface. With the ability to calculate a wide range of molecular descriptors, LSBD allows chemists to improve the accuracy of their predictive models with efficiency and ease.

Within a single run researchers can process a ligand set using one or more of the five theoretical methods encompassed by LSBD. Prime MM-GBSA and MacroModel eMBrAcE can rank-order ligands, LiaisonScore can be used to estimate absolute ligand binding affinity, while descriptors can be generated for linear interaction (LIA) or extended linear response (ELR) predictions. With the aid of Strike, Schrödinger's statistics module, descriptors are readily correlated with experimental values to predict absolute binding affinities.

LSBD's straightforward interface allows calculations to be set up with just a few clicks. Computed energetic terms and structural descriptors are compiled in a single structure file, making it trivial to pipeline results to other programs or share results with colleagues. The LSBD computational workflow is driven by a single keyword-pair input file, which makes it easy to modify, extend, or repeat calculations in the future.

 

Using the LSBD interface shown above, researchers can easily set up calculations to generate descriptors with the following programs:

• MacroModel: Ligand binding affinities can be rank-ordered or predicted with the aid of MacroModel eMBrACe calculations.
• Prime: LSBD provides an intuitive interface for running Prime MM-GBSA calculations.
• Liaison: Linear interaction (LIA) descriptors can be calculated with Liaison. Additionally, the LiaisonScore empirical scoring function provides an estimate of ligand binding affinity.
• QikProp: With the ability to predict over 35 pharmaceutically relevant ADME properties, QikProp can be used in conjunction with other descriptors from LSBD to generate extended linear response (ELR) models that predict ligand binding energies.