Pharmacophore-based compound modeling, virtual screening, and affinity profiling has become a popular and successful in silico technique for supporting medicinal chemistry. The advanced molecular design tool LigandScout [1] has been developed to successfully address one of the most important issues in virtual screening: Successfully enhancing early enrichment while maintaining high computational speed as well as ease of use, as shown by reference studies. [2,3]
As an extension of the static pharmacophore approach, we lately have focused on incorporating dynamic effects of ligand protein binding into our automated interaction determination process. Our Common Hits Approach (CHA) [4] uses multiple coordinate sets saved during MD simulations. Pharmacophore models with the same pharmacophore features are pooled and virtual screening runs are then performed with every representative pharmacophore model resulting in a consensus hit list. The recently developed GRAIL (GRids of phArmacophore Interaction fieLds) [5] method combines the advantages of traditional grid-based approaches for the identification of interaction sites with the power of the pharmacophore concept: A reduced pharmacophore abstraction of the target system enables the computation of all relevant interaction grid maps in short amounts of time. This allows one to extend the utility of a grid-based method for the analysis of large amounts of coordinate sets obtained by long-time MD simulations. In the NeuroDeRisk project [6], we utilized these new developments, together with machine learning methods for adding quantitative pharmacophore feature weighting [7] to predict potential neurotoxic effects of drug candidates. In the peptide field, we have used our technology to predict peptide bond replacement by triazole positions in order to maintain affinity while enhancing stability.
In addition to the algorithms made available in the LigandScout package [8], we develop the Chemical Data Processing Toolkit (CDPKit) [9] and make it available as an freely available open source chemoinformatics toolkit implemented in C++.
References:
[1] Wolber G, Langer T, J Chem Inf Model. 2005; 45(1):160.
[2] Karaboga AS, et al., J Chem Inf Model. 2013; 53(3):1043.
[3] Gallego RA, et al., J Med Chem. 2023; 66(7):4888.
[4] Wieder M, et al., J Chem Inf Model. 2017; 57(2):365.
[5] Schuetz DA, et al., J Chem Theory Comput. 2018; 14:4958.
[6] NeuroDeRisk IMI2 JU has received funding under grant agreement No 821528
[7] Kohlbacher SM, et al., J Cheminform 2021;13:57
[8] LigandScout, 4.5, Inte:Ligand GmbH, Vienna/AT (https://www.inteligand.com).
[9] Seidel T, Chemical Data Processing Toolkit (https://github.com/molinfo-vienna/CDPKit)