Study On Disease Feature Extraction And Inference Model Based On Immune Cell Receptor Diversity

In recent years,the science and technology and the social environment have undergone rapid changes.The increasingly rich material culture has greatly satisfied people's needs.With the aging population,people have given more attention to their health.The human immune system provides a powerful backing for the human body from viruses,bacteria and other invaders,and immune cells play a decisive role in the human immune system.With the application of immunotherapy in the treatment of cancer,human beings have great enthusiasm for exploring the correlation between immune cells and diseases.Studying the diversity of immune cell receptors can find out the pathogenesis of related diseases,and it is of practical significance and application value for people to provide disease prevention and early diagnosis of diseases by using relevant information.The main contents of this study are as follows: The high-throughput sequence-based sequencing algorithm for genome-wide assembly is proposed,and a high-throughput sequence-based sequencing algorithm for immunohistochemistry is proposed.The results of the stitching algorithm are compared and evaluated.In the aspect of immune cell receptor characteristics,a feature extraction model that maintains the original data format features is established and applied to the feature data set for feature extraction.In the modeling of disease reasoning model,the extracted feature set is applied to the inference model to analyze the disease inference results.The paper mainly contains the following three parts:In the high-throughput sequence connecting algorithm,a new high-throughput sequence-based connecting algorithm is proposed,which is evaluated by simulated sequencing data and showed high connection rate and accuracy.In the feature extraction section,the generated 29 million-dimensional features are effectively reduced to 6000.Both ELASTIC NET,LASSO and RIDGE feature extraction algorithms can extract the feature set information from the sample set as much as possible.At the same time,the information contained in the classification and discrimination information is very close,and the results using the classification reasoning model are also very close.In particular,ELASTIC NET preserves the original sample information in the most relevant set of features and is also the most efficient feature extraction algorithm.In the part of inference model research,the sample data is designed experimentally based on the distribution characteristics of samples,and the inference model is applied to the experimental data set.The research on inference model theory and actual operation show that adaptive boosting tree,random forest and support vector machine show similar results,because of there is a certain mechanism of immune cell receptors,random forest performance is better.