5. Conclusion
In the present work, we have described and made available FabSim, a new approach to reduce the complexity of tasks associated with computational research. We illustrate the scope and flexibility of FabSim by presenting its application in three diverse domains where it has been, and continues to be, used to simplify computational tasks and to improve their reproducibility. Our use cases in bloodflow modelling, materials modelling and binding affinity calculation provide evidence that FabSim benefits computational research on a generic level. In the area of brain bloodflow, we have described how FabSim can be used to do systematic benchmarking, to execute an ensemble of multiscale simulations, and to simplify the deployment of HemeLB on remote machines. In the nanomaterials area, we have shown how FabSim automates iterative parameterization of coarse-grained potentials, and allows us to systematically model the self-assembly of layered composite materials with chemical specificity. FabSim has also been applied to streamline the calculation of ligand–protein binding affinities through our Binding Affinity Calculator, allowing users to automatically launch ensemble computations, and thereby controlling uncertainties and producing reproducible results. In all cases, FabSim assists in the curation of run data by furnishing information about the job specification and the environment variables. The extent to which FabSim has been applied and adapted in these three domains serves to demonstrate its flexibility and ease of adoption. Indeed, using FabSim we have been able to publish our research findings in leading scientific journals in each domain.
