Skip to content

Large scale, in silico interaction analyses of SARS-CoV-2 nucleocapsid protein variants against human cytokines.

License

Notifications You must be signed in to change notification settings

tuplexyz/SARS-CoV-2_N-Cytokine_Docking

Repository files navigation

Human Cytokine and Coronavirus Nucleocapsid Protein Interactivity Using Large-Scale Virtual Screens

Phillip J. Tomezsko, Colby T. Ford, Avery E. Meyer, Adam M. Michaleas, Rafael Jaimes III
MIT Lincoln Lab and Tuple

Large scale, in silico interaction analyses of SARS-CoV-2 nucleocapsid protein variants against human cytokines.

Journal Article

About This Repository

In this repository, you'll find both the HADDOCK and AlphaFold2-Multimer results presented in the above paper. Plus, we have included the logic for preparing the various experiments and submitting them to an high-performance computing (HPC) environment.

HADDOCK 2.4 Analyses

HADDOCK 2.4 was run to dock 1,088 combinations of 64 human cytokines × 17 N proteins. We refer to these combinations as "experiments".

Ambiguous Interaction Restraints (AIRs) were defined using a random selection of 20% of the surface residues on the input proteins. This was to reduce our input bias surrounding the docking site while reducing computation time.

HADDOCK generates 200 PDB complexes for each experiment in each of its three iterations:

  1. Rigid body docking ($it_0$)
  2. Semi-flexible refinement ($it_1$)
  3. Solvent refinement in water ($it_w$)

In the final water refinement step, the HADDOCK system will cluster the complexes and generate cluster-level metrics.

From these cluster-level metrics, we select the best cluster based on the lowest van der Waals energy. Then, from this best cluster, we select the best (representative) PDB file as the one with the lowest van der Waals energy.

Results

Resources

AlphaFold2 Multimer Analyses

AlphaFold2-Multimer was also run to dock the same 1,088 "experiments" (combinations of 64 human cytokines × 17 N proteins).

Results

Resources

Data Explorer App

We have provided a basic data explorer that allows for the generation of figures and the viewing of the PDB complexes. This application is written in Streamlit. To run the application locally, use the following commands:

via Python (Terminal or Command Prompt):

cd vis_app/
# python -m pip install -r requirements.txt
streamlit run Protein_Viewer.py

via Docker:

docker build -t n_cyto_app -f vis_app/Dockerfile .
docker run -p 8501:8501 --name n_cyto_app -d n_cyto_app

## To run in an interactive shell:
# docker run --rm -it --entrypoint bash n_cyto_app
# docker exec -it n_cyto_app /bin/bash

Then, navigate to http://localhost:8501 in your web browser.

vis_app