With the help of control engineering, it is possible to achieve a better understanding of the dynamics of biological systems and predict their behavior through modeling. In the last two decades, biological research has greatly benefited from the advancement of technology and integration in Omics data at different levels, including genomics, transcriptomics, and proteomics. The purpose of this thesis is to design a model for integrate Omics data and also to predict communication between Omics layers and also, a tool should be designed to use the results obtained by the model. Because the cause of some diseases such as cancer and diabetes is not only one gene, and for its diagnosis and treatment, there is a need for communication between layers and molecular networks. Recommender systems also have an activity similar to the purpose of this research, which is to predict new connections, and for this reason, the models designed in this field were examined. On the other hand, because by displaying the data examined in this thesis in the form of a graph, it helps a lot to understand the data, the LightGCN model, which is based on the graph convolutional network and deep learning, was chosen because this model has higher speed and accuracy than other models in this field. The changes needed to meet the needs of this research were made in this model. The implementation and programming of the model designed in this research was based on the PyTorch library and the Python programming language. Because it had a higher speed compared to other existing libraries in this field. Also, to increase its speed and reduce the volume of calculations to a large extent, the COO format was used to display the matrices used in this research. In this thesis, two different categories of data are used, which include relationships in two categories, miRNA and mRNA, Drug and Target. These data sets only express the existence of edges, and for training the model, there is a need for data that also express the absence of edges, for which the negative sampling method was used. With the help of the knowledge obtained from the discovery of edges and existing connections, with the designed model, it is possible to predict other edges; which provides useful information to researchers and doctors. The performance of the model for ridge prediction in the available data set has been eva‎luated by the criteria of AUC, F1 score, sensitivity and accuracy. In order to be able to use the results obtained by the model and new predicted connections, an online tool for detecting and predicting the connection between miRNA and mRNA in three separate parts, which able to cover all the needs, with the help of the Django framework is designed to help monitor progress and help treat diseases and can be used by doctors and patients for their specific needs. It was also tried to make this tool have a suitable user interface and with the help of special search techniques, the response speed of the tool increased a lot.

جعفر قيصري

يوسف قيصري

ايمان ايزدي نجف آبادي , حامد جلالي بيدگلي