| Breast cancer,as one of the most common cancers with the highest mortality rate,involves involves a large number of oncogenes and tumor suppressor genes,so studying the mechanism of action of breast cancer-related genes is important for the diagnosis and treatment of breast cancer.Among them the MDM2 protein is over-expressed as a carcinogen in breast cancer to promote tumor development,while p53 protein is considered a tumor suppressor that can block mitogenic signaling and plays a key role in the cellular response to DNA damage.MDM2 is an important regulator of p53 and the p53-MDM2 regulatory relationship is important for p53 expression,and the development of valuable drugs holds promise as a new frontier for the treatment of breast cancer through targeted therapy against the p53-MDM2 regulatory loop or by directly blocking its regulation.MDM2 is an important regulator of p53,and the p53-MDM2 regulatory relationship is important for p53 expression,and the the development of valuable drugs holds promise as a new frontier for the treatment of breast cancer through targeted therapy against the p53-MDM2 regulatory loop or by directly blocking its regulation.In order to understand the pathogenesis of breast cancer,we investigate the dynamic behavior of p53-MDM2 mediated by mi RNA-29 a.At the same time,as the development of information technology and the improvement of computing speed,artificial intelligence and other popular information technology have made breakthroughs and advances,the combination of artificial intelligence technology and advances,the combination of artificial intelligence technology and disease treatment has become a hot topic of current research.For breast cancer,treatments include surgery,radiation therapy,chemotherapy and molecular targeted therapy.Among them,the advent of molecularly targeted therapeutics has provided a new direction for efficient and low-toxic treatment of breast cancer.The development of targeted drugs that target signals associated with the development of barest cancer has become a focus of research in breast cancer treatment.Finding compounds with favorable bioactivity,metabolic dynamics and safety,including absorption,distribution,metabolism,excretion and toxicity(ADMET properties for short),is a long and challenging task in the treatment of breast cancer,and this paper to improve the prediction ability of ADMET properties of breast cancer drugs in the screening process,molecular descriptor data and machine learning algorithm were used in this study to investigate the ADMET properties of compounds.In addition,breast cancer as a heterogeneous disease defined by molecular types and subtypes,in which one triple negative breast cancer of unmet medical need is the most aggressive and lethal of the breast cancer type of breast cancer,and the expression of three most targeted biomarkers in the treatment of triple breast cancer is absent,and targeted therapy is not feasible for triple negative breast cancer.However,with the development of high-throughput sequencing technology and genomic research,cancer research at gene expression level and cancer diagnosis and treatment technology is promoted,so the utilization of gene expression data and machine learning algorithms for triple negative breast cancer classification prediction is also studied in this paper The main research efforts are as follows:The research background,significance and current status of p53-MDM2 signaling pathway in breast cancer were investigated,and introduced that principle of machine learn method in the prediction of ADMET property of anti-breast cancer drug and triple negative breast cancer prediction based on machine learning,commonly used data preprocessing method,feature selection,evaluation index,etc.To analyze the pathogenesis of breast cancer from a dynamical point of view,the kinetic analysis of the p53-MDM2 model mediated by mi RNA-29 a has been studied through analytical and numerical simulations to provide a reference for the study of the pathogenesis and therapeutic targets of breast cancer.In this paper,we consider the effect of mi RNA-29 a and construct a p53-MDM2 gene regulatory network mediated by mi RNA-29 a,based on the p53-MDM2 model established by previous works.A new mathematical model has been proposed and its dynamical properties have been studied.The dynamic information contained in the data is mined and analyzed.The genetic behavior of the disease onset process can be further understood by the stability analysis of the positive equilibrium point and the analysis of the influence of the mi RNA-29 a correlation parameter on the system.Using compound data from the ChEMBL database as a research subject,a breast cancer drug screening model based on compound molecular descriptors and ensemble learning algorithm was proposed,and mainly investigated the ADMET properties of compounds.First,forty feature variables that significantly affect the properties of ADMET were selected from the molecular descriptor data of the compound.Second,according to the stacking idea of ensemble learning algorithm,a TLSA(Two Level Stacking Algorithm)model is established,in which Random Forest,Extra Tree,Gradient Boosting Decision Tree,Decision Tree,Bagging,Ada Boost and K-nearest Neighbor algorithms were used as the first level for TLSA model,and Logistic Regression or Support Vector Machine were used as the second level for TLSA model to construct property classification model.Final,setup comparison experiments verify that the TLSA model performs well in predicting ADMET properties.A breast cancer classification model was proposed by using breast cancer RNAsequence data and clinical data from TCGA database as the study subjects.First,differential expression analysis was performed on the gene expression data,then the Random Forest algorithm was used to select the features of the gene expression data,and the Attention Mechanism was used to assign the weights to the selected features.The construction of the Attention Mechanism Optimized Random Forest model is described in detail,and a comparison experiment is designed to verify that the constructed model is an effective classification method and can improve triple negative detection. |
PDF Full Text Request