Method for predicting mutagenicity of chemicals through machine learning algorithm

Abstract

The invention belongs to the field of ecological risk evaluation test strategies, and discloses a method for predicting mutagenicity of chemicals through a machine learning algorithm. On the basis ofthe known compound molecular structure, the mutagenicity of the compound can be quickly and efficiently predicted by calculating the molecular fingerprint and applying the constructed method. The method is simple and quick, is low in cost, and can save manpower, material resources and financial resources required by experiment testing. The construction process of the method comprises the followingsteps: collecting chemical mutagenicity data; preprocessing the data; calculating a molecular fingerprint; selecting a machine learning algorithm and training a model; selecting indexes such as accuracy to evaluate the model; characterizing the application domain; in the construction method, inputting molecules to be detected, and outputting mutagenicity of the molecules to be detected. The prediction model established by the method has good fitting capability, robustness and prediction capability, can effectively predict mutagenicity of the chemicals, provides necessary basic data for risk evaluation and management of the chemicals, and has important significance.

Description

technical field [0001] The invention relates to a method for predicting the mutagenicity of chemicals by establishing a QSAR model, and belongs to the field of ecological risk assessment test strategies. Background technique [0002] Mutagenicity refers to the ability to induce genetic damage. Genetic material can be changed in several ways: gene mutation; chromosomal aberration; chromosome number change. Gene mutation refers to the sudden and heritable variation of genomic DNA molecules. These include substitutions of base pairs as well as additions or deletions of base pairs. Chromosomal aberrations are changes in the structure of chromosomes. Therefore, mutagens that can cause mutations can be divided into two categories. One is a mutagen that acts directly on DNA, and the other is a mutagen that acts on chromosome replication or division. [0003] There are different mutagenicity detection methods for mutagens with different modes of action. For mutagens acting on ...

AI Technical Summary

Problems solved by technology

These deficiencies are mainly reflected in the following aspects: First, when previous studies used calculation methods to predict the mutagenicity of chemicals, the training data used for modeling were mostly single experimental data or a combination of two experimental data, without coverage An experiment that detects all types of mutagenesis, resulting in a predicted result representing whether the chemical is mutagenic only on DNA or only on chromosomes

Prediction results are one-sided

Second, most of the prediction models established in the past did not characterize the application domains used by the models, resulting in large deviations in the prediction of the mutagenicity of some chemicals in the process of using the models

Images