RESUMO
To predict interactions between human and viral proteins, we combine evolutionary sequence profile features with a Siamese convolutional neural network (CNN) architecture and a multi-layer perceptron (MLP). Our architecture outperforms various feature encodings-based machine learning and state-of-the-art prediction methods. As our main contribution, we introduce two types of transfer learning methods (i.e., frozen type and fine-tuning type) that reliably predict interactions in a target human-virus domain based on training in a source human-virus domain, by retraining CNN layers. Our transfer learning strategies can effectively apply prior knowledge transfer from large source dataset/task to small target dataset/task to improve prediction performance. Finally, we utilize the frozen type of transfer learning to predict human-SARS-CoV-2 PPIs, indicating that our predictions are topologically and functionally similar to experimentally known interactions. Source code and datasets are available at https://github.com/XiaodiYangCAU/TransPPI/.
RESUMO
The outbreak of 2019-nCoV in the central Chinese city of Wuhan at the end of 2019 poses unprecedent public health challenges to both China and the rest world1. The new coronavirus shares high sequence identity to SARS-CoV and a newly identified bat coronavirus2. While bats may be the reservoir host for various coronaviruses, whether 2019-nCoV has other hosts is still ambiguous. In this study, one coronavirus isolated from Malayan pangolins showed 100%, 98.2%, 96.7% and 90.4% amino acid identity with 2019-nCoV in the E, M, N and S genes, respectively. In particular, the receptor-binding domain of the S protein of the Pangolin-CoV is virtually identical to that of 2019-nCoV, with one amino acid difference. Comparison of available genomes suggests 2019-nCoV might have originated from the recombination of a Pangolin-CoV-like virus with a Bat-CoV-RaTG13-like virus. Infected pangolins showed clinical signs and histopathological changes, and the circulating antibodies reacted with the S protein of 2019-nCoV. The isolation of a coronavirus that is highly related to 2019-nCoV in the pangolins suggests that these animals have the potential to act as the intermediate host of 2019-nCoV. The newly identified coronavirus in the most-trafficked mammal could represent a continuous threat to public health if wildlife trade is not effectively controlled.
RESUMO
High-throughput transcriptomics technologies have been widely used to study plant transcriptional reprogramming during the process of plant defense responses, and a large quantity of gene expression data have been accumulated in public repositories. However, utilization of these data is often hampered by the lack of standard metadata annotation. In this study, we curated 2444 public pathogenesis-related gene expression samples from the model plant Arabidopsis and three major crops (maize, rice, and wheat). We organized the data into a user-friendly database termed as PlaD. Currently, PlaD contains three key features. First, it provides large-scale curated data related to plant defense responses, including gene expression and gene functional annotation data. Second, it provides the visualization of condition-specific expression profiles. Third, it allows users to search co-regulated genes under the infections of various pathogens. Using PlaD, we conducted a large-scale transcriptome analysis to explore the global landscape of gene expression in the curated data. We found that only a small fraction of genes were differentially expressed under multiple conditions, which might be explained by their tendency of having more network connections and shorter network distances in gene networks. Collectively, we hope that PlaD can serve as an important and comprehensive knowledgebase to the community of plant sciences, providing insightful clues to better understand the molecular mechanisms underlying plant immune responses. PlaD is freely available at http://systbio.cau.edu.cn/plad/index.php or http://zzdlab.com/plad/index.php.
