Kieran Alden, postdoc in the Kreft Lab, has recently spoken at the British Socity of Immunology Mathematical Modelling Affinity Group Meeting, held at Microsoft Research in Cambridge. This event, held between the 8th and 9th May 2013, aimed to bring together scientists working in the areas of experimental and theoretical immunology, to discuss current challenges on lymphocyte development, receptor diversity, lineage commitment, cellular signalling and host-pathogen interactions, and how modelling and mathematical approaches could and have been applied to these problems
Kieran's talk was titled "Building Confidence in Biological Simulations as Tools for Scientific Research". The talk raised the issue of how mathematical and computational modellers determine how confident they are that their simulation adequately represents the biological simulation it has been built to represent. The talk covered three strategies that could be employed in relation to simulation confidence: good simulation design, argument-based validation, and methods within the Spartan Statistical toolkit. Confidence in simulation results is an important issue for the field of modelling and simulation, and remains one of Kieran's key research interests (see here for information on a workshop that Kieran will be co-chairing with regard to this interest)
This talk was summarised by the following abstract:
Computational models of the developing gut have been presented and discussed at previous meetings of this affinity group. This research coincided with the recent explosion of interest in the gut micro-environment, driven by a realisation of the profound effects the microbiota may have on the health of the host. Growing interest in the gut will most likely lead to a rise in animal experimentation where the microbial environment is manipulated, making efforts to provide alternative and complementary experimental techniques more urgent. We have recently been awarded NC3R funding to develop computational tools that model the molecular interactions within the microbiota, with the aim of reducing or replacing the need for particular animal experiments. Yet only five years ago this funding body found that just 11% of surveyed researchers agreed that computer simulation could one day accurately represent a whole host and thus be useful in their research. For our aim to become reality, we need to establish confidence in the simulation as a robust scientific tool. Any simulation is an abstract representation of the system it has been designed to capture, and thus it can be unclear whether the emergent behaviour observed in simulation is an artefact of implementation, or is actually representative of that captured system. In this talk I will draw from both our previously presented lymphoid tissue development model and our latest gut microbiota study to suggest strategies that aim to build confidence in results and hypotheses generated through simulation, including model design frameworks and statistical analysis techniques.