Introduction to Desmond

Desmond is an explicit-solvent molecular dynamics program developed by D. E. Shaw Research, created with an emphasis on accuracy, speed, and scalability. The molecular dynamics engine can be used for a variety of applications, and is a key component of many drug discovery, and materials science workflows.

A description of Desmond was published, along with performance data, as part of the conference proceedings of the ACM/IEEE Conference on SuperComputing 2006 (SC06) [1]. While developing Desmond, D. E. Shaw Research has introduced and extended a number of scientific algorithms, including new parallelization strategies and numerical techniques, some of which have been published [2–5].

For installation and configuration instructions, including required hardware and settings for queueing systems, see the Installation Guide and Desmond.

Synergy with the Schrödinger Suite

Desmond exists as a part of the larger Schrödinger suite, which contains a number of powerful tools to support various scientific challenges. Below are sources of synergy for Desmond across the Schrodinger suite:

• The Protein Preparation Workflow Panel, LigPrep (ligand structure) and Epik (ligand protonation state) preparation tools can be used to ensure that the structures provided to Desmond are chemically correct. Such careful system preparation often represents a crucial step prior to initiating a molecular dynamics simulation.

• MacroModel can be used to perform a quick check on input structures, to ensure Desmond will be able to meaningfully run on the structure. This can be done using a Current Energy calculation, with the Solvent set to None. If the structure is problematic, Maestro and MacroModel often provide useful diagnostics for what might be wrong.

• Prime can be used to create homology models for use in simulations and to repair protein structures.

• Glide can be used to generate relevant poses within protein binding sites for use in simulations. Desmond in turn can be used to thermally relax, refine, and sample conformations related to the docked poses.

• Desmond can be used to sample protein structures prior to performing docking calculations with Glide.

• AutoQSAR can be used to generate statistical models from the results of simulations.

• SiteMap can be used to identify potential binding sites from simulation results.

Schrodinger also offers a number of applications that are extensions of Desmond. These are generally advanced, but commonly used, workflows that require molecular dynamics simulations and include:

• FEP+, which uses free-energy perturbation theory to produce accurate relative binding free energies for a series of congeneric ligands to a protein receptor.

• WaterMap, which analyzes specially designed Desmond simulations to characterize the thermodynamics of water in protein binding sites.

• Various workflows in the Materials Science suite. Desmond is used for simulations calculating bulk properties of materials, including elastic constants, stress/strain relationships, diffusion coefficients, viscosity, persistence length, and thermophysical properties. See Bulk Properties in the Materials Science Suite for more information. The Materials Science suite also provides tools for preparing structures and running complex sequences of simulations, including coarse-grained simulations. See Molecular Dynamics in the Materials Science Suite for more information.