The Universal Immune System Simulator (UISS) is a computational framework based on agent-based modelling (ABM) paradigm that has been specifically developed for simulating the immune system behaviour in presence of diseases and treatments. It has a long history of development, ranging from its initial applications into the field of tumor immunology and then moving towards wide disease modelling scenarios such as influenza, Multiple Sclerosis and atherosclerosis. Recently, inside the STriTuVaD H2020 EU project, it has been specialized to simulate tuberculosis dynamics and its interaction with the immune system, including the efficacy of the combined action of various treatments such as isoniazid and novel vaccines. TB simulation entitles large scale (e.g., tissue to organ scale) simulations over a wide digital population cohort. The computational costs of running large scale simulations are prohibitive using traditional forms of CPU simulation. This paper considers the use of parallel to gpu-based computing approaches via an agent-based domain independent complex systems simulator, FLAME GPU. Integration of FLAME GPU with UISS enables the simulation of larger, more complex problem domains. The combined UISS-FLAMEGPU simulator provides vastly increased performance characteristics for large problems, with a speedup of 4.22x for a typical tuberculosis model simulating 128 microlitres. FLAME GPU abstracts away a significant portion of the normal programming that would be required to effectively parallelise a model of this complexity. Adaptations were made to increase performance, such as message mutation and parallelisation of certain algorithms.

UISS-GPU: Accelerated In-Silico Tuberculosis Vaccine Trials Using FLAME GPU

Pappalardo F.;Russo G.;
2022-01-01

Abstract

The Universal Immune System Simulator (UISS) is a computational framework based on agent-based modelling (ABM) paradigm that has been specifically developed for simulating the immune system behaviour in presence of diseases and treatments. It has a long history of development, ranging from its initial applications into the field of tumor immunology and then moving towards wide disease modelling scenarios such as influenza, Multiple Sclerosis and atherosclerosis. Recently, inside the STriTuVaD H2020 EU project, it has been specialized to simulate tuberculosis dynamics and its interaction with the immune system, including the efficacy of the combined action of various treatments such as isoniazid and novel vaccines. TB simulation entitles large scale (e.g., tissue to organ scale) simulations over a wide digital population cohort. The computational costs of running large scale simulations are prohibitive using traditional forms of CPU simulation. This paper considers the use of parallel to gpu-based computing approaches via an agent-based domain independent complex systems simulator, FLAME GPU. Integration of FLAME GPU with UISS enables the simulation of larger, more complex problem domains. The combined UISS-FLAMEGPU simulator provides vastly increased performance characteristics for large problems, with a speedup of 4.22x for a typical tuberculosis model simulating 128 microlitres. FLAME GPU abstracts away a significant portion of the normal programming that would be required to effectively parallelise a model of this complexity. Adaptations were made to increase performance, such as message mutation and parallelisation of certain algorithms.
2022
978-1-6654-6819-0
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.11769/549365
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact