[paper] Identification of SARS-CoV-2–induced pathways reveals drug repurposing strategies

Identification of SARS-CoV-2–induced pathways reveals drug repurposing strategies (2021)

- Science Advances, https://doi.org/10.1126/sciadv.abh3032

- bioRxiv, https://doi.org/10.1101/2020.08.24.265496

- Presentation : https://youtu.be/dqyzbC5ZSZA (Korean)

- Presentation : https://youtu.be/SE3dGRKp5s0 (English)

- Github : https://github.com/wchwang/Method_Pancorona

 

Fig. 1.

Method

SARS-CoV-2 와 직접적으로 관련있는 protein 데이터(cause)와 시간에 따라 protein expression 변화 데이터(consequence)는 연관이 있을 거라는 가정하에 (2% overlap between DIP and DEP)

1. DIP과 DEP의 SARS-Cov-2-induced protein (SIP) network(STRING v11)를 hidden layer로 구성

2. key pathway를 분석(applied multiple network algorithms, 1000 permutation tests with P-value < 0.01)

3. pathway를 inhibition하는 약물 확인

 

-  6h : 332 DIP and 64 DEP

- 24h : 332 DIP and 164 DEP

 

Result

- Artificial Neural Network를 통해 148개의 pathway(Reactome pathway enrichment analysis using gprofiler2)를 9개의 cluster(a self-organizing map, k-means clustering algorhtim with Davies-Bouldin Index)로 나누고, Mechanisms of Action(searched COVID-19-related literature)의 2 category로 나누었다. 그리고 각 pathway에 속한 drug이 몇개 인지 확인.

 

- Precision : the proportion of drug targets that are annotated to the pathway.

- Recall : the proportion of the pathway gene set that the drug targets recover.

Fig. 3

- 승인된 약물 중 관련있는 200개 약물을 찾았고, 이 중 40개는 이미 COVID-19 clinical trial에 사용되고 있으며, 30개는 COVID-19 치료 후보로 논문에 보고된 것이다.

- Uniprot Keyword enrichment test에서 200개 약물은 1573 protein에 targeted, 이 중 30 protein은 8 또는 그 이상의 약물에 targeted.

 

- Cellular assays를 통한 5개 약물의 validation을 진행.

- Initial screening으로 monkey Vero E6 cell line을 사용. Proguanil과 Sulfasalazine가 noticeable 세포 독성을 보이지 않음.

 

Discussion

Fig. 5. Schematics depicting the pathways mediating NO production that are targeted by the five tested drugs. The black boxes indicate key proteins in SIP network, and those targeted by the five drugs are highlighted in red color. Sulfasalazine and proguanil target proteins in both pathways that directly and indirectly (via NADP production) affect NO production.

실험을 진행한 5개의 약물 중, Proguanil과 Sulfasalazine 약물은 두 pathway를 inhibition으로 확인.

One limitation of our analyses is the use of undirected protein–protein network analysis, meaning it is unknown whether the predicted drug inhibits or accelerates coronavirus infection. For example, flucytosine was predicted to be associated with viral replication in this study and in our previously published study of viral replication, flucytosine accelerated rather than inhibited virus replication. This limitation can be solved by using a disease-specific directional network, which could be generated using time-series transcription data, phosphorylation data and a human protein interaction network, such as a regulatory network and a protein–protein network. If the methodology we have described in this paper was used on a disease-specific directional network, it could be used to predict drugs associated only with the suppression of virus infection.

 

Data set

- DIP, https://www.nature.com/articles/s41586-020-2286-9#Sec36

- DEP, https://www.nature.com/articles/s41586-020-2332-7#Sec23

- Protein : significantly up- or -down-regulated by two-sided, unpaired Student's test with equal variance assumed, P < 0.05, |log2FC| > 0.5.

- PPI confidence score > medium (0.4) : STRING v11, https://string-db.org/

- Approved drugs for repurposing : ChEMBL, DrugBank (v5.1), STITCH (v5.0, confidence score > 0.9), Network-based prediction of drug combinations.

- Reactome pathway (15 May 2020) over-representation analysis (ORA) for the target proteins of the drugs using gprofiler2 (Fisher's exact test, P-value < 0.05)

- Anatomical Therapeutic Chemical (ATC) code, https://www.whocc.no/atc_ddd_index/, https://www.who.int/tools/atc-ddd-toolkit/atc-classification

 

Network algorithm for Key pathway

Graph Algorithms: Practical Examples in Apache Spark and Neo4j

- Randomwalk with Restart

- Eigenvector centrality

- Degree centrality

- Betweenness centrality

 

 

 

Reference

- Direct Interacting Protein data from A SARS-CoV-2 protein interaction map reveals targets for drug repurposing (2020), https://doi.org/10.1038/s41586-020-2286-9 : 직접적으로 관련있는 protein을 질량분석기를 사용하여 얻음.

- Differentially Expressed Protein data from Proteomics of SARS-CoV-2-infected host cells reveals therapy targets, https://doi.org/10.1038/s41586-020-2332-7 : 시간에 따라 protein expression의 변화를 질량분석기를 사용하여 얻음.

- DISEASES: Text mining and data integration of disease–gene associations (2015), https://doi.org/10.1016/j.ymeth.2014.11.020, https://youtu.be/_2rfSxBSFnc

- DISEASES 2.0: a weekly updated database of disease–gene associations from text mining and data integration, https://doi.org/10.1101/2021.12.07.471296

- REST API of Enrichr, https://maayanlab.cloud/Enrichr/

- Circos, http://circos.ca/

- https://clinicaltrials.gov/

- Interplay between Cellular Metabolism and the DNA Damage Response in Cancer (2020), https://doi.org/10.3390/cancers12082051

- Viral infection linked to m6A alterations in host mRNAs (2020), https://doi.org/10.1038/s41580-019-0202-7

- Cell Cycle Perturbations Induced by Infection with the Coronavirus Infectious Bronchitis Virus and Their Effect on Virus Replication (2020), https://doi.org/10.1128/JVI.80.8.4147-4156.2006

- The self-organizing map (1990), https://doi.org/10.1109/5.58325