With the development of the epidemic, researchers urgently need to be able to sort out and systems integration knowledge novel coronavirus, the acceleration mechanism research and development of new antiviral drugs crown virus. For example, what the virus belongs to the family's new crown virus is an important attribute of the new virus crown of what the relationship between antiviral drugs target the virus is, and what kinship virus. This knowledge exists in many of the latest knowledge or the published literature, the researchers are difficult to retrieve and use the system.
To solve this problem, Huawei's cloud agent medical team, Huawei cloud voice semantic innovation Lab, Professor of Computer Science, Zhejiang University Chen Huajun jointly form a joint team base in Zhejiang University to build the first edition of the early maps on Taxonomy of Viruses, has added three new scientific knowledge maps, full coverage of the basic properties of the new crown virus, providing a powerful tool for the study of viruses and antiviral research and development research staff.
The first version of the new crown research spectrum from the perspective of biological classification of the virus to NCBI data in the United States Center for Biotechnology Information Taxonomy plate is constructed based on a virus of the family network: Taxonomy of Viruses map. The pattern with the larger, the number of instances in which more than 200,000, the number of triples is reached + 1.9 million. But it does not introduce additional information other than the family relationship and classification, with some limitations.
Therefore, second edition, by a joint team concepts combing virus, genes, proteins, drugs, etc., about the novel coronavirus, antiviral drugs work in such areas. Departure information genes, proteins, and the nucleotide sequence of the host from viruses and the like, based on the new crown NCBI database virus (SARS-CoV-2, the former name: 2019-nCoV) data, to build a new crown basic map information v1.0 . From antivirals, Human Protein, Virus Protein, the starting host information, and the like based on the data DrugBank database, constructs a map of antiviral v1.0. Finally, given the AIDS epidemic is an important field of research directions, real-time new crown virus sequence data joint team to Gisaid global influenza database provided, combined nextstrain for analysis and correlation algorithm new crown viral genome epidemiology, sorting out a short mutant virus network connection - viral genetic relationship map v1.0, which includes geography, time information, the degree of variation, the mutant gene protein and the like, to provide support for the subsequent fusion studies (virus Origin) and the map.
The new crown basic information Atlas v1.0
express SARS-CoV-2 represents a relationship between the expression of its genes, such as SARS-CoV-2 expression of membrane protein gene (M) / Envelope protein gene (E) / Nucleoprotein gene (N) and other types of genes.
Translation translate showing the relationship between gene and protein, e.g. membrane protein gene (M) SARS-CoV-2 can be translated membrane protein.
produce relationship expressed SARS-CoV-2 can produce the corresponding protein.
host used to express SARS-CoV-2, and the relationship between the host, the host entity is currently maps are mainly human and vertebrates.
Basic information spectrum diagram of the new crown
Spectrum antiviral drugs v1.0
antiviral effect relationship for the virus used to indicate a certain effect, for example Ritonavir and Abacavir Human immunodeficiency virus 1 antiviral effect, this relationship can be viewed by antiviral drugs and the corresponding contact between the virus.
Mining can be used to produce the relationship between the virus and they express the relationship between the expression of proteins, and further explore the linkage between the different viral proteins of the same type, and further found that the relationship between the two virus indirectly, between a virus and then by a relation of antiviral drugs , but also found some indirect relationship antiviral drugs and other viruses that can help find more drugs may have a role for a certain type of viral proteins.
binding interaction and relationships can be used to discover the interaction between viral proteins and host proteins, such as the HIV1 NP (nucleoprotein) may be in the Human binding of HYOUl; can be replenished after the relationship between the target and HYOUl certain drugs, these can be studied further HIV1 interaction between the drug and the NP (nucleoprotein).
In terms of attributes, from DrugBank joint team excavated the following information (trade name, chemical formula, molecular weight, indication, etc.), as an attribute drug entity, so researchers can more intuitive understanding of some important information about a drug. After the team will further improve the key entities to add more important information.
Antiviral schematic map
Virus genetic relationship Atlas v1.0
Knowledge Graph has Strain strain, Branch branch, Country national, State and City urban area these types of entities, where the geographic location and connection Strain strains identified by from_country, from_division and from_location. Country, State, City interconnected network configuration.
For strains Strain properties, including AA mutation, nucleotide mutations, and a branch on the difference rate information, etc., shown in the figures in the table. Branch node and wherein variability data are difference data rate is relatively mutate_from_branch indicated, and the upper layer is relatively Branch Branch, so the final sum Strain mutation should be mutated entire directory tree.
Schematic map of the virus genetic relationship
Potential applications of the new crown research map
Potential applications Xinguan research map as follows: 1. Prediction of biological classification new virus variant virus predictive 2. 3. 4. The thermal stability of the prediction the predicted virus pathogenic virus susceptible population portion 5. 6. Prediction Prediction virus virus 7. a pharmaceutical symptoms may lead to potential treatment of new or old drug 9. the predicted protein may interact with viral proteins by transmission of the virus 8. prediction, the discovery of new protein targets 10. the targets for a new protein, 11. traceable virus drug discovery, analysis and prediction of virus mutation.
Building knowledge map is usually a lengthy and laborious process, in the construction of the new crown research map, the joint team used Huawei cloud knowledge map knowledge map construction services were end to end, and the use of Huawei's cloud-stop AI ModelArts development and intelligent platform management drawn from the literature of the new entity relationship, in just a week's time to build a content-rich crown new research maps, which greatly improves the efficiency and accuracy, the next team will be more joint drawn from the latest literature to update their knowledge to the knowledge Graph.