| Background and purpose:Colorectal cancer(CRC)is a common malignancy in the digestive system.The common histological types are adenocarcinoma,adenosquamous carcinoma,spindle cell carcinoma,squamous cell carcinoma,and undifferentiated carcinoma.Among them,colorectal adenocarcinoma(COADREAD)is the most common,accounting for more than 90.Although the current diagnosis and treatment of colorectal cancer continues to advance,the mortality rate of advanced CRC is still high.According to reports,the 5-year survival rate of patients with early colorectal cancer is close to 90%,while the 5-year survival rate of patients with advanced metastatic colorectal cancer is less than 10%.Therefore,the early diagnosis and treatment of CRC are particularly important for patients.The Guideline for colorectal cancer from the National Comprehensive Cancer Network(NCCN)suggests that screening for colorectal cancer be stool occult blood tests combined with colonoscopy.Although the high specificity and sensitivity of colonoscopy,it is a minimally invasive operation,more expensive,complex operation of the surgery,it is not suitable for population screening.Therefore,there is a need to further seek new methods of high sensitivity,high specificity,non-invasiveness,and ease of operation for early screening,diagnosis and follow-up of CRC after treatment to improve the prognosis of CRC patients.This study studied the relevant clinical data of COADREAD and mRNA in TCGA,and used artificial neural network to establish a model for predicting the prognosis of CRC.The relationship between mRNA and COADREAD survival time was analyzed to provide basis for further research on the occurrence and development of COADREAD.Artificial Neural Network(ANN)is a marginal discipline.It is the result of the interactive development of computer science,information science and medicine in recent years.The basis of understanding is the organizational structure and operating mechanism of the human brain.The essence of ANN is artificial intelligence research,which uses the computer’s powerful computing capability to simulate the process of animal neural network information transmission.Its essence is an ordinary statistical model that can be used to solve multi-factor and nonlinear problems.At present,artificial neural networks have been successfully applied in many fields,and have obtained many betterresults.Neural networks are very powerful in processing uncertain information and have very good robustness.In recent years,the number of bioinformatics papers using neural networks has continued to grow.So far,the cancer is the most complicated disease in the world.Each cancer has its own molecular characteristics.Therefore,it is very important for human health to understand the genetic changes of each type of cancer.The Cancer Genome Atlas(TCGA)program is funded by the U.S.government,was established by the National Cancer Institute(NCI)and the National Human Genome Research Institute(NHGRI)in cooperation with many institutions in the United States and Europe.It is the world’s largest project on the genome project,and its purpose is to apply the latest genomic analysis technology to deeply understand the changes of malignant tumor genes and discover new ways to prevent,diagnose and treat tumors.The network platform for the research results of this project is the TCGA network platform,which is currently the largest and most commonly used public resource database in cancer research.It allows researchers to search for and download tumor-related data for analysis with no pay.TCGA collected tumor tissues from more than 11,000 patients and matched it with normal tissues.A comprehensive,multi-dimensional map of key gene changes has beendrawn for 33 cancers.The Generic Data Commons Data Portal(GDC)is TCGA’s powerful data-driven platform that links to external analysis tools such as cBioportal and Firehose.The Firehose platform is used in this study.To date,TCGA has provided data on genomic sequences,RNA sequences,tumor miRNA and mRNA expression,and methylation of more than 30 different tumor samples.In 2012,TCGA analyzed the mRNA expression of 224 cases of COADREAD.In this study,the Firehose platform provided 621 colorectal adenocarcinoma samples and 18041 mRNAs per sample.Data and methods:1.Download the clinical data of colorectal adenocarcinoma on the TCGA platform,and select 573 samples of age,sex,primary site,AJCC pathological stage,and survival time,and apply the Kaplan-Meier method for single factor long-term(this study refers to five-year survival rate)survival analysis,P <0.05 for statistical significance.2.Download clinical data of colorectal adenocarcinoma and related mRNAs on the TCGA platform,and select 400 mRNAs that may be correlated with survival of colorectal adenocarcinoma with statistical significance at P<0.05 and Q<0.3.The first 10 mRNAs with the smallest P value were tested by logrank to construct an artificial neural network model forpredicting the prognosis of colorectal adenocarcinoma.SPSS 22.0 was used for analysis.Result:1.The long-term survival rate of colorectal adenocarcinoma has nothing to do with age,gender,primary site,and is related to the stage of AJCC pathology at the time of initial diagnosis.2.The same mRNA has different expression levels in different samples,and the expression level is different,and there are differences in survival.3.Based on the data provided by the TCGA network platform,the first 10 mRNAs with the smallest P value were tested using the logrank test,and a model predicting the prognosis of the COADREAD sample was constructed using an artificial neural network.Conclusion:1.The long-term survival rate of colorectal adenocarcinoma has nothing to do with age,sex,primary site,and pathological stage at the time of initial diagnosis.The prognosis of patients with stage III and stage IV AJCC pathology is poor.2.The same mRNA has different expression levels in different samples.The amount of expression is different,the survival is different.3.The expression of part of the mRNA is related to the prognosis of colorectal adenocarcinoma samples.4.Theartificial neural network model of mRNA can be used to predict the prognosis of COADREAD samples. |
