Organic fluorescent small molecule optical property prediction method based on deep neural network

Abstract

The invention provides an organic fluorescent small molecule optical property prediction method based on a deep neural network. According to the method, a new organic fluorescent small molecule database is established, molecular descriptors and molecular fingerprints are adopted to extract molecular information, and a multi-layer neural network and a convolutional neural network are input for deep learning training to obtain an organic fluorescent small molecule optical property prediction model; the characteristic information of the to-be-predicted organic fluorescent micromolecules and the experimental solvent thereof are input into the trained organic fluorescent micromolecule optical property prediction model so as to predict the optical property of the to-be-predicted organic fluorescent micromolecule. The method can accurately predict the optical properties (the average relative error is less than 5%) of the organic fluorescent small molecules, so that the development efficiency of the organic fluorescent small molecules is improved.

Description

technical field [0001] The invention relates to the cross field of computer science and chemical fluorescent probes, in particular to a method for predicting the optical properties of small organic fluorescent molecules based on deep neural networks. Background technique [0002] With the development of life sciences and research needs, fluorescence imaging technology has gradually become one of the important means of qualitative and quantitative detection. Among them, organic fluorescent small molecules have the characteristics of small structure, easy chemical transformation, and excellent optical properties. Their design and synthesis have always been one of the research hotspots in the field of fluorescence imaging. At present, we have a certain understanding of the relationship between the structure and optical properties of organic fluorescent small molecules, but there are still limited reports that can accurately predict their optical properties based on molecular st...

AI Technical Summary

Problems solved by technology

At present, the development of most new organic fluorescent molecules is still based on synthesizing a series of compounds at the same time, so as to examine their optical properties one by one, and screen out molecules with ideal fluorescent performance, which has the limitations of manpower, time cost and resource investment, and low success rate.

Images