Fig. 5

The flowchart of WMGHMDA model to predict potential human microbe-disease associations. The first step is constructing a heterogeneous network by connecting the microbe similarity network where the microbe similarity is obtained by combining the Gaussian kernel microbe similarity with the microbe functional similarity, the disease similarity network where the disease similarity is obtained by combining the Gaussian kernel disease similarity with the disease semantic similarity, and the known microbe-disease association network. The second step is iteratively executing Weighted Meta-Graph search algorithm on the heterogeneous network to calculate the scores of the microbe-disease pairs. Finally, prioritizing candidate microbes for diseases according to their scores