RESUMEN
The phoD-harboring bacterial community is responsible for organic phosphorus (P) mineralization in soil and is important for understanding the interactions between arbuscular mycorrhizal (AM) fungi and phosphate-solubilizing bacteria (PSB) at the community level for organic P turnover. However, current understanding of the phoD-harboring bacterial community associated with AM fungal hyphae responses to organic P levels remains incomplete. Here, two-compartment microcosms were used to explore the response of the phoD-harboring bacterial community in the hyphosphere to organic P levels by high-throughput sequencing. Extraradical hyphae of Funneliformis mosseae enriched the phoD-harboring bacterial community and organic P levels significantly altered the composition of the phoD-harboring bacterial community in the Funneliformis mosseae hyphosphere. The relative abundance of dominant families Pseudomonadaceae and Burkholderiaceae was significantly different among organic P treatments and were positively correlated with alkaline phosphatase activity and available P concentration in the hyphosphere. Furthermore, phytin addition significantly decreased the abundance of the phoD gene, and the latter was significantly and negatively correlated with available P concentration. These findings not only improve the understanding of how organic P influences the phoD-harboring bacterial community but also provide a new insight into AM fungus-PSB interactions at the community level to drive organic P turnover in soil.
Asunto(s)
Hongos , Micorrizas , Fósforo , Humanos , Microbiología del Suelo , Bacterias/genética , Fosfatos , SueloRESUMEN
The aim of the study is to build a tongue image intelligent analysis "end-to-end" deep learning network based on a tongue diagnosis image of traditional Chinese medicine. The tongue target region in the original image was segmented by the UNet tongue segmentation model at the front end of the network. After segmentation, the feature vector of the tongue target region was extracted by the ResNet network, and then the blood pressure on the day of shooting was fused with the feature vector extracted by the ResNet network through the convolution operation method to complete the extraction of two groups of data of tongue feature and fusion feature. Based on analyzing the data of blood pressure, tongue image, and their fusion at the end of the network, four regression analysis methods were used to predict the stage mean value. After training, the model is tested with the test set data, and the test results are evaluated with mean absolute error (MAE). The prediction error of the model based on the fusion data of tongue image and blood pressure on the day of shooting was lower than that of the other two data modes. The UNet tongue segmentation model combined with the ResNet network can realize the automatic extraction of tongue image features. The extracted features combined with machine learning modeling can be used to explore the complex hierarchical mathematical association between tongue image and clinical data. The experimental results show that the multimodal data fusion method is an important way to mine the clinical value of the TCM tongue image.