Molecular screening method based on deep learning technology qualitative and quantitative model

Abstract

The invention discloses a molecular screening method based on a deep learning technology qualitative and quantitative model (EMIM), which comprises the following steps: converting collected molecules into SMILES, and then converting the SMILES into pretreatment of qualitative and quantitative descriptors; carrying out hash mapping, variance threshold and Pearson's correlation coefficient feature engineering on qualitative and quantitative descriptors; setting a qualitative and quantitative model according to input data processed by feature engineering, using a Sigmoid function as final output, optimally setting parameters of the qualitative and quantitative model through a back propagation optimization algorithm to perform performance evaluation, and performing iterative training to obtain a model with high verification precision; enabling a to-be-predicted molecule to be subjected to preprocessing and feature engineering and then inputting the to-be-predicted molecule into the highest verification precision qualitative and quantitative model (EMIM) for prediction screening, and obtaining a prediction result of the to-be-predicted molecule. According to the method, more efficient and accurate screening of new molecules is realized, and the limitation of traditional screening is solved.

Description

technical field [0001] The present invention relates to the intersection fields of deep learning and physics, chemistry, and materials, and in particular relates to a qualitative and quantitative model built using a deep learning neural network framework, which can be used for molecular screening. Background technique [0002] Molecular descriptors are the characterization of molecular structure sub-fragments or physical and chemical properties, and can be divided into qualitative descriptors and quantitative descriptors. Qualitative descriptors generally refer to molecular fingerprints, which can convert chemical molecules into bit strings containing only 0 and 1. Quantitative descriptors describe molecules based on molecular composition (number of hydrogen bond donors, number of benzene rings), physical and chemical properties (topological polar surface area, octanol-water partition coefficient) and experimental data information (ultraviolet spectrum, solvent ratio), etc. ...

AI Technical Summary

Problems solved by technology

Therefore, the application of machine learning and deep learning in screening new molecules is limited to a certain extent.

Images