HPC-enabled multiscale simulation helps uncover mechanistic insights of the SARS-CoV-2 infection

 A Use Case by

Short description

To be able to use all the potential of future Exascale systems, researchers will need an efficient and sustainable infrastructure to support the development of personalised medicine. PerMedCoE is building such an infrastructure by upscaling tools that are used in personalised medicine projects, such as, to translate the consequences of omics information into actionable molecular disease models. Multiscale modelling prove useful in integrating mechanisms with different time and space scales. We have enabled the use in HPC of PhysiCell, a multiscale modelling framework and one of PerMedCoE’s core tools, to expand its scope and study the dissemination and infection of the SARS-CoV-2 virus by incorporating cell- and pathway-specific Boolean models to detail the interactions of virus, drugs and human cells. These Boolean models are simulated using MaBoSS, another PerMedCoE’s core tool, allowing for the study of genetic and environmental perturbations. PerMedCoE is collaborating with the COVID-19 Disease Map and the PC4COVID initiatives and introducing leading-edge technologies to discover biomarkers and actionable therapeutic targets that will help against the COVID-19 pandemic.

Results & Achievements

BSC has successfully implemented MPI in the multiscale modelling and the use of shared memory has been an enabling technology to have more complex, bigger multiscale simulations. This opens the possibility, for instance, of having simulations of real-size tumours of a billion cells.

Even though this PerMedCoE use case is in its preliminary stages, we have identified two genes, one from the SARS-CoV-2 virus and the other from the human epithelial cells, whose inactivation allows for the evasion of Apoptosis or programmed cell death. Thus, these two genes are strong candidates as therapeutic targets in COVID-19 infection.

Objectives

In this use case, our main aim is to uncover mechanistic insights that could help in the fight against SARS-CoV-2. For this we use Boolean models of signalling pathways, agent-based models for populations of cells and the communication among virus, epithelial host cells and immune cells. The multiscale scope of the modelling and the breadth and depth of the HPC-enabled simulation using MPI facilitate the uncovering of therapeutic targets. In addition, the organisation of this work in building blocks and pipelines allows for an optimised orchestration and distribution of the different tasks in the project.

Finally, a global aim of PerMedCoE is to serve as an example for the upscaling of other personalised medicine tools that are part of PerMedCoE’s observatory.
Figure 1: Top-level overview of the multiscale simulation workflow that integrates virus infection, epithelial host cell demise and different immune cells’ response.
Figure 1: Top-level overview of the multiscale simulation workflow that integrates virus infection, epithelial host cell demise and different immune cells’ response.
Figure 2: Simulation of mutants and heterogeneous cell populations. Our framework can simulate wild type cells (A) and study how knock outs, such as FADD’s, alter the model’s behavior (B). Our framework also allows for studying different proportion of cells with an inhibited virus’ M protein, such as 50% of M knock out cells (C) and 95% of M knock out cells (D).
Figure 2: Simulation of mutants and heterogeneous cell populations. Our framework can simulate wild type cells (A) and study how knock outs, such as FADD’s, alter the model’s behavior (B). Our framework also allows for studying different proportion of cells with an inhibited virus’ M protein, such as 50% of M knock out cells (C) and 95% of M knock out cells (D).

Technologies

PhysiCell
MaBoSS
PhysiBoSS
nanoHUB
PC4COVID initiative
COVID-19 Disease Map initiative

Use Case Owner

Barcelona Supercomputing Center-Centro Nacional de Supercomputación (BSC-CNS)

Collaborating Institutions

Institut Curie
The PerMedCoE consortium

more