Desmond

Desmond is a GPU-powered, high-performance molecular dynamics (MD) engine developed by Schrödinger, designed for simulating a wide range of biological and chemical systems. It is suited for researchers working with small proteins, viral capsids, protein-ligand complexes, small molecules in mixed solvents, organic solids, and synthetic macromolecular complexes. Desmond delivers exceptional scalability, throughput, and scientific accuracy, making it a trusted tool for computational drug discovery and life sciences research.

As an official NVIDIA partner, Schrödinger has optimized Desmond for NVIDIA GPU technology, enabling computing speeds up to 100x faster than single-CPU performance on general-purpose GPUs (GPGPU). The engine runs efficiently on commodity Linux clusters with both typical and high-end network configurations.

Key Performance and Accuracy Benefits

• GPU-accelerated performance: Achieves exceptional throughput on commodity Linux clusters and improves computing speed by 100x on GPGPU compared to single CPU processing.
• Superior numerical accuracy: Built with a focus on numerical accuracy, stability, and rigor, enabling simulation of large-scale features ranging from nanometer to micron size over time scales from picoseconds to microseconds.
• Trusted energetics: Provides a robust framework for calculating energies and forces for atomistic force field models, compatible with chemistries commonly used in biomolecular research.
• Realistic simulations: Performs explicit solvent simulations with periodic boundary conditions, with careful attention to long-range electrostatics calculations; supports modeling of protein and nucleic acid systems with explicit lipid membranes.

Usability and Interface Features

• Easy-to-use interface: Provides intelligent default settings and allows rapid setup of computational experiments through an intuitive interface.
• Automated simulation setup: Supports automated workflows including system building, analysis tools, and force field assignment.
• Powerful analysis tools: Enables visualization and examination of computed results within the Maestro modeling environment, which connects to a comprehensive suite of tools spanning quantum mechanics to machine learning.

Key Application Areas

• Mixed Solvent Molecular Dynamics (MxMD): Improves cryptic pocket identification through enhanced sampling. Features a dedicated interface for simplified setup, analysis, and customizable visualization of cryptic binding pockets on protein surfaces.
• Unbinding Kinetics: Characterizes ligand-receptor interactions through unbinding kinetics analysis. Visualizes unbinding pathways using enhanced sampling methods to identify and optimize promising lead compounds based on their dissociation rates.
• Simulation of complex protein solutions and preparation of complex protein systems for MD workflows.
• Building and analyzing complex lipid bilayers and embedding membrane proteins.
• Enzyme engineering and investigation of mutation effects on enzyme stability and ligand binding.
• Generating ternary complex structures to enable rational design of targeted protein degraders (PROTACs).
• Predicting drug residence times from unbinding kinetics simulations on kinase systems.
• Calculation of thin plane shear viscosity and friction coefficients.
• Oligonucleotide modeling and computational target analysis for nucleic acids in drug discovery.

Available Tutorials and Learning Resources

• Step-by-step tutorials covering MD simulation setup, lipid bilayer construction, enzyme engineering, unbinding kinetics, and mixed-solvent MD.
• Online certification courses providing hands-on molecular modeling training with access to Schrödinger software.
• Comprehensive documentation, quick start guides, videos, and support resources.
• Peer-reviewed publications demonstrating Desmond's application across life science and materials science research.

Desmond integrates seamlessly within the broader Schrödinger platform, working alongside related products including the Maestro modeling environment, OPLS4 and OPLS5 force fields, FEP+ for free energy calculations, IFD-MD for ligand binding mode prediction, and Virtual Cluster for scalable cloud-based simulation. This tight integration enables end-to-end computational workflows from system preparation through advanced analysis within a single unified environment.