Rational drug design is developed based on the targets that interact with drugs, such as enzymes, receptors, ionic liquids, viruses, nucleic acids, polysaccharides, etc., to find and design reasonable drug molecules. The structure of the drug at the molecular level or even the electronic level can be determined by comprehensively and accurately understanding the related mechanism. In the rational design of new drugs, natural products with new structures and precise pharmacological effects or known synthetic drugs are often used as the lead, and the chemical structure is modified and simplified through drug design to make it more suitable for the treatment of diseases. The development of computer science and the emergence of high-performance computers have greatly improved the speed and accuracy of data calculation and analysis. On this basis, rational design of novel drugs based on computational simulation has been involved in all aspects of drug research as a practical tool, and has become one of the core technologies of innovative drug research.
Figure 1. General pipeline for rational drug design by integrating QSAR methods with receptor-based computational method. (Yang, G. F.; Huang, X. 2006)
Alfa Chemistry provides easy-to-use models and computational tools for rational design of novel drugs. Our rapid and high-quality services are as follow:
In the early stages of rational drug design, molecules with new structures can be constructed by combining fragments of existing compounds or using optimization algorithms. Machine learning, especially deep learning, has also been applied to drug discovery, such as predicting the properties and activities of compounds and their interactions with protein targets. Alfa Chemistry supports a diversity of commonly used generative models, such as recurrent neural networks (RNN), autoencoders (AE), generative adversarial networks (GAN), transformers, and hybrid models that combine deep generative models and reinforcement learning.
1. RNN model: After a large number of SMILES string training, the RNN model can be used to generate a new effective SMILES not included in the original data set. The RNN model is therefore considered as a molecular structure generation model.
2. AE: We apply AE to de novo drug design in which AE is used to study the probability distribution of the data set to generate samples similar to but different from the data set.
3. GAN: Our teams train the GAN to obtain new samples from the generator.
4. Transformers: The key function is its attention mechanism which fully considers the long-range dependence in the sequence.
Based on the space coordinates of the atoms of a given molecular system, a reasonable molecular system structure is obtained through multiple iterations of numerical algorithms. The force field uses multiple typical structural parameters and forces to describe the changes in the structure, helping to design a more rational drug molecule.
1. Force field applicable to macromolecules: AMBER, OPLS/AMBER, CHARMM, CHARMm, glorious mode, ECEPP (free energy force field).
2. Force field suitable for small molecule positions: MM3, CFF95, MMFF94, UFF.
3. Energy minimization: Stochastic descent method (SD), conjugate gradient method (CONJ), arbitrary step length approximation method, Newton-Raphson method.
4. Conformation analysis
1) Molecular dynamics: Our scientists first optimize the molecule geometrically , and then perform dynamics simulation based on it. At Alfa Chemistry, we use two sampling methods for molecular dynamics: simulated annealing kinetics and high temperature quenching kinetics.
2) Random conformation search: We mainly use Monte Carlo method and Metropolis sampling method.
3) Systematic conformation search: In general, our teams apply grid search to systematically searching the conformational space of molecules to find the smallest points on the potential energy surface.
4) Other methods: Genetic algorithm and distance geometric method are also available for conformation analysis.
The chemical nature of the drug is determined by the basic structural characteristics of its peripheral electrons. The activity indicators of quantum mechanics in drug design include: orbital energy, charge density, bond level, delocalization energy, atomic self-polarizability and electrostatic potential.
1. We apply ab initio method to accurately calculate the weak interaction energy in molecular systems using Mɸller-Plesset (MP) method, density functional theory (DFT) and Hartree-Fock.
2. We also use a series of simplified ab initio calculation methods such as valence electron (VE) ab initio calculation, floating ball Gaussian orbit (FSGO) method, molecular fragment (MF) method and simulated ab initio molecular orbit (SAMO) method to quickly and effectively determine the geometric configuration.
3. Our experts also apply multiple models to predict solvation free energies involving a large number of solvents with high accuracy.
We use a series of structural features such as physical and chemical parameters, molecular topological parameters, quantum chemical indicators and structural fragmentation index to study the relationship with the biological activity of compounds. Our groups apply cluster analysis (CA), principal component analysis (PCA), nonlinear transformation (NLM), artificial neural network (ANN) and factor analysis (FA) for pattern recognition.
Our rational design of novel drugs molecules services remarkably reduce the cost, promote further experiments, and accelerate the process of drug design for customers worldwide. Our personalized and all-around services will satisfy your innovative study demands. If you are interested in our services, please don't hesitate to contact us. We are glad to cooperate with you and witness your success!