MultiMod, flexible management for multi-scale multi-approach models in biology

Even the simplest living organisms perform a huge number of different processes, which are interconnected in complex ways to ensure that the organism responds appropriately to its environment. One of the ways of ensuring that we really understand how these processes fit together is to build quantitative mathematical models of them which can be simulated using computers. This approach is known as "Computational Systems Biology".

As in most domains of science, databases are essential to allow the proverbial "standing on the shoulders of giants". BioModels provides access to quantitative biology models that have been published in the scientific literature, and verified to be accurate. For instance, in order to develop a quantitative model of cell tumorigenesis, one may choose a suitable model of cell-cycle, and attempt to merge it with models of relevant cell signalling pathways such as the MAP kinase cascade. Since its creation at EMBL-EBI in 2005 - and supported by the BBSRC since 2008 - BioModels has undergone an exponential growth to become the worldwide reference for quantitative models of biological processes. Deposition of models upon publication is advised by several hundred scientific journals, and the resource receives around a million page requests a year from around 65 000 distinct users.

Over the recent years, mathematical models of biological processes have become larger and more complex. To be able to represent more accurately living structures, multi-scale models are developed, with processes taking place at different scales described by components coupled together. Because of the nature of the processes as well as the experimental information available, a variety of modelling approaches are use. In 2012, a landmark publication described the first such model of a bacterium (Mycoplasma genitalium) that accounts for all of its components and their interactions. Multi-scale models of organs and organisms (for instance the BBSRC-funded model of Arabidopsis) are now available. Those models must be made available, regardless of their structure, the modelling approach used and the formats they are encoded with.

The MultiMod project, developed in collaboration by the Babraham Institute and EMBL-EBI, will extend BioModels coverage considerably, to support a large variety of mathematical models developed in the biosciences. This will be achieved by accepting more file formats, improving support for standard formats and expanding the information distributed (for instance including necessary experimental datasets, simulation descriptions etc). We will also expand the spectrum of models we verify by reproducing published results. Such an increase in scope will be accompanied by an improvement of the software and hardware infrastructure underlying the resource. To make the resource more useful we will develop better browsing, search and download capabilities. Finally, we will improve and further develop our documentation and training offer, including online and in person tutorial and courses.

We expect that MultiMod will improve the user experience significantly, increase the number of users and the communities using the resource, and make it more useful for the UK biosciences.

Models, formal representations of biological systems, are an essential part of modern, quantitative biology, making the essential step from descriptive to predictive research. BioModels is the globally leading curated database for systems biology models, in 2014 serving >1,000,000 pages to >60,000 distinct users. The MultiMod project will further develop BioModels, extending its scope to new requirements of the community both technically and in terms of curation procedures, enhancing its infrastructure to cope with increased load and model complexity, and optimising its outreach impact. Based on the JUMMP platform developed in the recently completed BBSRC BioModels grant [BB/J019305/1], we will increase support for SBML Level 3 Packages, working from released packages for logic, constraint-based, reaction-diffusion, and modular models, to packages under community development. Based on the JUMMP-based model repository we developed for the IMI project DDMoRe, we will implement support for Systems Pharmacology models. Using the COMBINE archive format, we will develop flexible support for non-standard, but heavily used additional formats like Matlab, and support inclusion of simulation descriptions and relevant datasets. The technical extensions to the scope of BioModels will be flanked by the extension of curation approaches in collaboration with the respective communities. The initial focus will be on Flux Balance Analysis and logical models. We will also further develop a range of core controlled vocabularies like SBO, KiSAO and MAMO. Beyond core infrastructure and interface improvements of browsing, search, and download, we will increase interactivity of the BioModels interface through improved direct online simulation capabilities, based on the Systems Biology Simulation Core Library, and optimise documentation and outreach through an integrated context-sensitive help system, online courses as part of the EMBL-EBI training portal, and YouTube or similar.

The impact of MultiMod will occur over different time scales, each involving different parts of the population and society: modellers and scientists using models, users of models to generate knowledge and goods, and larger impact on society.

The immediate impact of MultiMod will take place via direct users of BioModels. The use of models is growing in modern biosciences, as we move from the description of systems to mechanistic understanding and predictions. Exchanging well encoded, well annotated models greatly facilitates the activity of model builders and users. BioModels had a tremendous impact on modelling in systems biology, in particular when it comes to kinetic modelling of biochemical processes. It is expected that extending its scope will extend its impact to other communities. MultiMod will therefore save time, energy and money in any project using models to understand biological systems. Furthermore, models are knowledge integrators. However, building mathematical models goes further than data integration, providing functional testing that accelerates hypothesis validation and research. Multi-scale multi-approach models are also a key tool of 3Rs approaches. As such models grow, they will be more accurate replacements of animal testing, as numerical simulations replaced nuclear testing. Being able to easily access trusted versions of the model will be very important.

Outside of life sciences, mathematical modelling is omnipresent, and hardly any material or good is produced without one or several modelling steps. Having access to a comprehensive set of existing models for any given system will limit costs by reducing useless re-invention, and generate wealth by fostering new developments. Finding the right models, using them and comparing their results, will speed up the prediction, development and testing of a whole range of goods, whether drugs, bioengineered products, treatments etc. For instance, freely sharing pharmacometrics and systems pharmacology models will have a strong effect on drug discovery, whether to predict drugs or to determine treatments. Modelling is also becoming more important to study "pathway diseases". Precision medicine aims at understanding precisely the molecular anomaly causing a disease in a given patient, so as to choose the best drugs and regimens. This will increasingly rely on understanding mechanistically the effect of the anomaly and the drugs. Similarly, multi-scale models of tissues or plants will eventually affect the production of food. As bioengineering unfolds, producing biological material and tools to perturb them so will the use of models to improve this production.

Modelling a process is a key step to check that one understood it. Materials from BioModels are already used at different levels of higher education, from bachelor programmes to advanced postgraduate courses, to teach biological pathways or modelling chemical kinetics. Because of the variety of models available, MultiMod will extend this impact in various sub-disciplines. It will help train a generation of interdisciplinary scientists and engineers, who will shape the UK landscape in systems modelling and its application for the years to come.