Research On Neural Network For Uncertain Genotype Data And Economic Data

Genome-wide association analysis(GWAS)is one of the important analysis methods for people to study genetic diseases.By searching for the correlation between target genes and specific diseases,the genes with strong correlation are found from the loci.The current research on genotypes is mostly the assumption of regression or genetic models to study the association analysis of genotypes and diseases.In the study of uncertain genotypes,Z Kutalik et al.(2017)^[1]used accurate maximum likelihood estimation to perform association analysis on the data.Neural networks are rarely used for this kind of research.Due to the weak requirements of the neural network for the model,the specific model in real life is unknown.In the research,the model is usually assumed to be a linear model or other models,but this may be affect the actual fitting effect.In high-throughput sequencing,it is almost impossible to get all specific genotypes accurately,usually we only get the genotype probability.In this paper,neural network research is used to conduct gene association analysis,and specific analysis is carried out in the case of uncertain genotype.The research sequence of this paper is as follows:First,introduce genes and their related backgrounds,and introduce neural networks into data with uncertain genotypes;Second,the binary logistics model under the“maximum probability method”and“dosage”was simulated by a neural network,and it was found to have a good fitting effect.Use the weight of the neural network to find the more relevant pathogenic genes,and compare the advantages and disadvantages of the two methods;Third,in order to the research of the extended neural network on the time data of uncertain genotypes,through the deep feed-forward neural network(DNN)and the recurrent neural network(RNN),the price of the stock's opening price in the next day was studied,and the corresponding accuracy was given;Fourth,the full text is summarized and prospected,and the feasibility of neural networks on genetic time series data is discussed.