Optimization Of Vina Docking For TCM Based On Machine Learning

Malnutrition(PEW)is one of the most common complications in patients with chronic kidney disease(CKD)and is an independent risk factor for the prognosis of patients with CKD.Because of its complex etiology,the current lack of effective treatment.The traditional Chinese medicine prescription kidney failure decoction has good curative effect on CKD-PEW,but the exact mechanism is unclear.As the traditional Chinese medicine prescription more complex components,it is difficult to screen and interpretation of which the active ingredient and pharmacological mechanism is not conducive to the standardization of traditional Chinese medicine prescription and optimization.How to select the active components from the large number of small molecules is the bottleneck of TCM research.With the development of bioinformatics in recent years,great progress has been made in the analysis of large data.Among them,machine learning and biological data processing are the most potential biological information analysis methods.Machine learning algorithm can be obtained from the data automatically analyze the law,and use the law to predict unknown data algorithm.Because of the complexity of Chinese medicine prescription and mechanism,it is difficult to describe and predict biological phenomena with simple mathematical model,and machine learning is the best means to predict drug targets and determine pharmacodynamics.Therefore,using machine learning and other biological information technology as a means of analysis,in order to reveal the compound traditional Chinese medicine prescription drug efficacy substances and mechanism of action provides a strong method and technical support.Therefore,it is possible to predict the active components of the target by training the target relationship of known drugs.Therefore,it is possible to provide new ideas and strategies for elucidating and optimizing the mechanism of Chinese herbal compound by means of machine learning.In the present study,the potential targets and prescriptions of the prescriptions were screened by means of data analysis,and the results were validated by in vitro experiments.METHODS:The gene expression data of skeletal muscle PEW model were selected in GEO database.The SAM gene was used to screen the differentially expressed genes.The enrichment of GO and enrichment of KEGG based on different genes were used to predict PEW involved in biological function and signal pathway.The MisgDB Sub-enrichment prediction related transcription factor binding motif,thus screening the key signals and possible drug targets;from a number of Chinese medicine database to retrieve the chemical composition of the group;application of Vina program based on free energy for small molecules and target bulk and the active components were screened out by using four kinds of machine learning models based on molecular fingerprints.The effects of screening components on protein metabolic rate of skeletal muscle were detected by isotope labeling method,and the WB method was used to detect and detect the influence of screening components on the binding of the target gene to the promoter was detected by reporter gene assay.The effect of the selected components on the expression of key genes was detected by qPCR.The results showed that skeletal muscle PEW was involved in many signal transduction pathways,such as protein catabolism,and FoxO1 might be the key target.Based on the molecular docking of FoxO1,seventeen potential interacting molecules were screened out.Based on the application of fingerprinting and its learning model,the specificity of activity prediction could be improved to above 0.9.Combining with four kinds of machine learning models,we predicted that glycyrrhizin B(HB)Inhibitor.Compared with the model group,immunofluorescence showed that HB could significantly increase the radius of the myotubes(P<.05).Isotope labeling showed that HB could significantly reduce the rate of protein degradation in the PEW model(P<.05)(P<0.05).The results of luciferase reporter gene showed that HB could inhibit the promoter sequence of FoxO1 binding.The results of qPCR showed that HB could significantly inhibit the expression of Atogin-1 and MuRF-1 in the FoxO1 downstream atrophy-related genes Expression level.Conclusion:The molecular fingerprinting-based machine learning model can significantly improve the specificity of molecular docking.It is predicted that active ingredient HB can significantly improve the PEW status of skeletal muscle.The mechanism involves inhibition of FoxO1 Se257 phosphorylation and inhibition of downstream gene expression.