We highlight new features and enhancements to the Schrödinger software in 2013. A focus will be placed on docking (Glide), structure refinement (Prime), pKa prediction (Jaguar), molecular dynamics (Desmond), and biologics (BioLuminate). Highlights include a new peptide docking mode, an improved covalent docking method, physics-based membrane permeability model, protein linker design, and antibody humanization.
 

We show that combining results from multiple methods using a standard score (Z-score) can significantly improve results over any single method or averages of multiple methods. We use three different virtual screening methods (2D fingerprint similarity with Canvas, shape-based similarity with Phase Shape, and docking with Glide HTVS) to screen eleven targets. Database enrichments with data fusion were significantly better than any of the individual methods. The average enrichment for the entire set was improved by 25% over the top individual method and by 43% over the average of the individual methods using data fusion. This enhanced approach to combine data from diverse methods can readily be applied to improve database enrichments in virtual screening projects and will be available as a script in the upcoming 2013-1 software release.

Jaguar pKa is a computational package for pKa prediction based on density functional theory (DFT). For superior accuracy, the method must use two empirical parameters per each functional group. Consequently, the coverage of the chemical space becomes an important factor determining application to real-world systems. In this seminar, we will present a new hierarchical model for extending the coverage of the pKa prediction to essentially all functional groups. The pKa results obtained on a diverse set of drug-like molecules in water will be demonstrated. Additionally, we will discuss the remaining challenges of pKa prediction and propose directions for future progress.