Rapid estrogen activity screening method based on molecular surface point cloud

Abstract

A construction method of an estrogen activity prediction model, a screening method of estrogen activity, an electronic device and a computer readable storage medium. the construction method of the estrogen activity prediction model comprises: obtaining known chemical data with estrogen activity, the chemical data comprising initial three-dimensional structure information of chemicals; optimizing the initial three-dimensional structure information to obtain optimized three-dimensional structure information; based on the optimized three-dimensional structure information, obtaining a molecular surface point cloud matrix of the chemical; and taking the molecular surface point cloud matrix as an input training convolutional neural network model to obtain the estrogen activity prediction model. According to the deep artificial neural network model constructed by the invention, quantifiable structure parameters which are artificially defined do not need to be used as molecular descriptors, so that the time and computing resources of molecular descriptor calculation and descriptor selection are saved, and the requirement on computational chemistry basis is lower during application.

Description

technical field [0001] The invention relates to the technical field of environmental health risk assessment of chemicals, and more specifically relates to a rapid screening method for estrogen activity based on molecular surface point clouds. Background technique [0002] A large number of environmental chemicals have gradually been found to have estrogen-like activity. These chemicals can simulate the biological behavior of estrogen in the human body, thereby interfering with the normal function of the human endocrine system and causing adverse health effects on the human body. The endocrine disrupting effect of exogenous compounds, especially pollutants, has aroused widespread concern in society. In order to protect people from such potential risks, the government must conduct strict estrogen-like activity evaluation and production and application control of chemicals that can come into contact with the human body. However, compared with the thousands of chemicals in the ...

AI Technical Summary

Problems solved by technology

There are still some problems with this method: they cannot describe the three-dimensional structure information of the molecule, such as the orientation of the group and the bond length, etc.; secondly, the description of the atoms in the molecule is too simple, ignoring the influence of the atomic properties on the surrounding environment

[0004] In summary, although the quantitative structure-activity relationship mathematical prediction model based on traditional machine learning algorithms has greatly improved the process of chemical evaluation and rapid screening of properties, due to the limitation of available descriptors, it cannot be used in more complex systems. It is difficult to achieve sufficient prediction results; and the calculation and collection of descriptors requires a certain amount of time, computing resources, and a certain discipline foundation, which also limits the application of prediction models to a certain extent

Images