SARS-CoV-2 is a relatively new pathogen, and there is a very limited number of experimentally validated drugs which were shown to be effective against it. We manually curated a list of such compounds from a literature search (see Additional file 7). Special care was taken to include only compounds with experimental rather than predicted evidence and explainable mechanism of action. Preference was given to the compounds already included in clinical trials. This resulted in a list of 49 “positive” class compounds. Drugs were putatively classified into two subgroups—acting through cellular mechanisms directly against the virus (“Direct-Cell”) or having symptomatic effects, e.g. anti-inflammatory (“Symptomatic”). This classification is not strict as drugs may have overlapping functions. Finally, drugs were checked for direct target overlaps with the “Aggregated” set of COVID-19-related genes. Drugs with very few to no overlaps in the top 100 genes were marked as less “reliable”. This resulted in four subselections for the “positive” class compounds which were tried in the model parameter optimisation stage: (1) “Target_Cell”: all drugs acting directly on the host-viral interactome, 27 in total; (2) “Target_Cell_Strict”: same as above, but only the most “reliable” drugs included, 19 in total; (3) “Target_Cell_Sympt”: both symptomatic and host-viral interactome targeting drugs included, 49 in total; and (4) “Target_Cell_Sympt_Strict”: same as above, but only the most “reliable” drugs included, 28 in total.

For the “negative” class, both approved and experimental drugs from DrugBank were selected. The antiviral drugs designed to target specific viral protein targets (such as remdesivir, tenofovir and taribavirin) were designated as neutral (“0”) class and were excluded from the model calibration. This is because the primary objective here is to target SARS-CoV-2 host interactome networks rather than individual viral proteins.

All available compounds from DrugBank and FooDB which were not included in the “positive” and “negative” classes were not used at the model calibration and parameter optimisation stage. Six thousand five hundred ninety-three compounds formed the input “negative” class; however, depending upon the specific parametrisation settings, the final number of negative class compounds varied between 1181 and 4260 due to the drugs with no connections being automatically removed.

Note: The content above has been extracted from a research article, so it may not display correctly.



Q&A
Please log in to submit your questions online.
Your question will be posted on the Bio-101 website. We will send your questions to the authors of this protocol and Bio-protocol community members who are experienced with this method. you will be informed using the email address associated with your Bio-protocol account.



We use cookies on this site to enhance your user experience. By using our website, you are agreeing to allow the storage of cookies on your computer.