Drug design is based on computer chemistry, through computer simulation, calculation and budgeting of the relationship between drugs and receptor biomacromolecules, the method of designing and optimizing lead compounds. Computer-aided drug design is actually the optimization and design of lead compounds by simulating and calculating the interaction between receptors and ligands. Computer-aided drug design generally includes active site analysis, database search, and new drug design.
Figure 1. Computer-aided drug design.
The general principle of drug design is to first obtain the structure of the receptor macromolecule binding site through X-single crystal diffraction technology and other techniques, and use molecular simulation software to analyze the structural properties of the binding site, such as electrostatic field, hydrophobic field, and hydrogen bonding site. Point distribution and other information. Then use database search or new drug molecular design technology to identify molecules whose molecular shape and physical and chemical properties match the receptor site, synthesize and test the biological activity of these molecules. After several rounds of cycles, new lead compounds can be discovered . Therefore, computer-aided drug design generally includes active site analysis, database search, and new drug design.
Traditional drug design method
Artificial intelligence (AI) technologies, especially new methods such as deep generative models, have shown bright prospects in the de novo molecular design of new drug development, and also in other difficult issues such as hit-to-lead structural modification and compound drug-making optimization. Will play an increasingly important role. Through the use of statistical language models and transfer learning ideas in natural language processing, the docking is processed. In the construction of AI model, MedAI uses the long and short-term memory (LSTM) neural network model that is commonly used in natural language processing problems in current generative models. We can build an LSTM neural network composed of an LSTM layer, a Dropout layer, a TimeDistributed layer and a Softmax layer. As a typical Recurrent Neural Network (RNN) model, LSTM is suitable for dealing with the problem of using standard SMILES format to describe molecular structure generation, and its gated structure can also learn the characteristic properties of molecules when the SMILES formula is longer.
|Project name||Drug design service|
Feasibility assessment of new drug targets
|Product delivery mode||The simulation results provide you with the raw data and analysis results of molecular dynamics.|
MedAI have cooperated with many universities or research institutions and can consult industry experts for their opinions and suggestions on some complex issues. Our drug design team is a highly professional, efficient and well-established R&D team.
Tech Trends in Artificial Intelligence What's up, what's down, what lies ahead, only found in MedAI!