Abstract:
Understanding the microbe-disease relationship can not only reveal the pathogenesis of diseases, but also promote the diagnosis and prognosis of diseases.Based on the structured deep network embedding algorithm, a new method (NEMDA) is proposed to identify potential microbial-disease associations. First, by integrating the human microbe-disease association database (HMDAD) and Disbiome databases, the number of microbes and diseases as well as the known microbial-disease associations have been expanded. Next, the structural deep network is embedded to extract features on the microbe-disease bipartite network, and microbe functional similarity, microbe interaction profile similarity, disease semantic similarity and disease symptom similarity are introduced as the biological characteristics of microbes and diseases, respectively. Then, these three characteristics are combined to form the characteristics of the microorganism-disease pair, and the deep neural network model is used to make predictions. Finally, the performance of NEMDA is evaluated through five-fold cross validation and case analysis. Under five-fold cross validation, NEMDA performs well, and its prediction performance is higher than the six comparison methods of KATZMDA, NCPHMDA, LRLHMDA, PBHMDA, NTSHMDA and BRWMDA. The case studies of asthma, inflammatory bowel disease and colorectal cancer further show that NEMDA has good predictive performance. Therefore, NEMDA is an effective tool for predicting microbial-disease associations.
