Study on the mechanism of gene regulation network of flower primordia development in Arabidopsis based on recurrent neural network

Abstract:
Arabidopsis thaliana is a cruciferous model plant, and its research is of great significance. In this paper, based on Recurrent Neural Networks (RNNs), a gene regulatory network model of Arabidopsis flower primordium development was constructed to explore the dynamic gene regulation process of flower primordia development. Select experimentally validated regulatory networks, and select key genes in network. Using 0-5d expression data to build an RNN model. Then, the RNN model was used to simulate the quantitative regulatory relationship between key genes. In this study, the RNN model constructed based on the flower primordia developmental regulatory network can achieve the combined input of different gene expression levels at multiple time points and obtain a quantitative output, the expression of key genes. Further, after adding the irrelevant gene ATO into the regulatory network and introducing 10% Gaussian noise, the RNN model can effectively fit the regulatory relationship between genes and has good robustness.