TTConf (Powered by TetraBox)
Significant Speed-Up: Provides up to 24x faster results compared to state-of-the-art methods on datasets like CD25, BACE and Astex
High Accuracy: Delivers precise conformers with fewer function evaluations.
Supports multiple backends for energy evaluation: Force Fields, Semiempirical Electronic Structure Methods, Ab Initio Electronic Structure Methods
Benefits:- Fast and accurate screening of numerous local minima.
- Generation of comprehensive conformer ensembles.
- Utilizes data-agnostic tensor train optimization.
Prototyped in collaboration with Prof. Christoph Bannwarth & Christopher Żurek
TTBind (Powered by TetraBox)
Optimized Binding Positions: Utilizes tensor train optimization to find optimal ligand binding positions.
Automatic Binding Site Detection: Identifies binding sites without prior knowledge.
Supports multiple backends for energy evaluation: Force Fields, Semiempirical Electronic Structure Methods
Smart Optimization: Features intelligent internal coordinates to enhance optimization efficiency.
TTBind (Powered by TetraBox)
Enhanced Precision: Achieves 40% better minima than the GSO algorithm in LightDock, with comparable computational cost.
Focused on rigid-body protein-protein docking for improved performance.
Utilizes data-agnostic tensor train optimization
Supports multiple backends for energy evaluation: Force fields, Semiempirical electronic structure methods
Incredible Speed: Offers up to 400× speed-up compared to state-of-the-art CPU methods.
High Accuracy: Delivers precise energy calculations using ab initio methods.
Key Features:
- CUDA-Enabled quantum chemistry simulations
- Advanced Calculations: Closed-shell Hartree-Fock and DFT calculations
- Efficiency: Features RI-J and SnK methods for studying large molecules
- Functional Variety: Includes LDA and GGA functionals provided by LibXC
*Benchmark details: Coulomb matrix calculation using RI method, C60 molecule with cc-pvdz basis and cc-pvdz-jkfit auxiliary basis. Results are for 1x NVidia V100 vs single CPU core of Intel Xeon Scalable 8173M. Tests were run on GCP environment.
TQchem Tools
Single-Point Energy Calculations
• Custom-developed GPU-accelerated Ab initio HF and DFT methods
• Prediction of electronic properties of the molecule
• Supports multiple backends: force fields, semiempirical and ab initio methods
Geometry Optimization
• Global and local optimization with both gradient-free tensor train and gradient-based methods
• Optimization with constraints and restraints
• Optimization using internal coordinates
Internal Coordinates Scan
• Analysis of the energy landscapes w.r.t. change of an internal coordinate
• Manually refine conformers
• Select initial conformer for optimization
• Evaluate structure dependent properties
Dr. Roman Ellerbrock
Head of Applied Research, Chemistry
Dive Deeper with Our Research Publications
