Double-layer neural network algorithm used for high-precision energy calculation of organic molecular crystal structure

Abstract

The invention belongs to the organic molecular crystal structure technology field, and especially relates to a double-layer neural network algorithm used for high-precision energy calculation of an organic molecular crystal structure. The algorithm comprises the following steps of carrying out a first round of conventional crystal structure prediction; extracting all molecular conformations from an existing crystal and calculating energy; extracting all molecular dimers in a Van der Waals radius range of a central unit cell, and calculating intermolecular interaction energy; carrying out monomolecular conformation analysis, and constructing a convolutional neural network of monomolecular conformation energy; constructing the convolutional neural network of molecular dimer energy correction; and calculating total crystal energy. In the invention, precision of energy calculation in a drug molecular crystal structure prediction process is improved, and a calculation speed is maintained; and through rapid and accurate energy calculation, CSP is guided to rapidly find a truly stable crystal form on a correct potential energy surface.

Description

technical field [0001] The invention belongs to the technical field of organic molecular crystal structure prediction, and in particular relates to a double-layer neural network algorithm for high-precision energy calculation of organic molecular crystal structures. Background technique [0002] The characteristic of compounds that form different crystal structures is called polymorphism. The key physical and chemical properties of the compound itself, such as density, shape, solubility, dissolution rate, etc., are strongly affected by its crystal form. For pharmaceuticals, the crystal form can strongly affect the bioavailability of the drug and ultimately the therapeutic performance of the drug. Experimental polymorphic drug screening has become an essential part of the standard drug discovery process. In the experiment, people set the key crystallization parameters manually or with the help of robots, but it is difficult to obtain the correct crystallization conditions i...

AI Technical Summary

Problems solved by technology

[0004] For the first challenge, the completeness of crystal space sampling, it is usually done through large-scale crystal structure search, which will generate a large number of crystal structures and require a lot of energy calculations. For inorganic CSP, it is usually directly used The crystal energy is obtained by the quantum mechanical precision calculation method, but the organic molecular crystal has too many crystal structures that need to calculate the energy in the CSP process due to the high complexity of the system and the high dimension of the chemical space, which hinders the direct use For the application of the calculation method of quantum mechanical precision in the organic CSP scene, an alternative method is to use the classical mechanical method with low precision but fast calculation speed, but it is limited by the precision limit of the classical mechanical calculation method, this method often Make the potential energy surface description of structure prediction inaccurate

Images