Knowledge Transfer for Modelling at Organism Scale

Context and key challenges

To predict phenotypes, especially at cellular and tissular levels, constrained mathematical models have been developed for living organisms at different scales, ranging from the cell to the entire organism.

Such models are developed in three steps: the automated extraction of biological information and data from public databases; the reconstruction of biological networks (whether metabolic or cellular); and their enhancement through manual curation carried out by expert researchers.

This last step, which is widely dependent on human expertise and heterogeneous data, is currently a limiting factor, even for the modelling of relatively simple organisms such as yeasts or bacteria.

A major challenge when modelling organisms lies in the assessment of the extent to which models developed using a reference species (such as Arabidopsis in plant science, and mice or humans in the study of mammals) can be transferred to closely-related species that might bring agronomic or zootechnical benefits, and to define effective transfer strategies.

In this context, the purpose and originality of the KOMOD consortium is to establish a new multidisciplinary scientific community within INRAE, focused on the modelling of complex organisms and/or organisms that are of potential agronomic or zootechnical benefit. This community will be built around a common interest in the representation, formalization and encoding of knowledge, in order to make it easier to transfer and reconstruct biological processes, including metabolic networks. Its goal is to encourage convergence between the plant and animal research communities through the sharing of modelling methods and frameworks.

Goals and methodology

In building the consortium, the project leaders seek to coordinate and federate the modelling community, bringing together expertise in biology, bioinformatics and mathematics/modelling as a means to:

1. transfer the information integrated in models from one context to another (between different tissues, environmental conditions or organisms);
2. define strategies for transferring models to other biological systems;
3. identify, share and employ generic bioinformatic and biostatistical algorithms that allow model specialization or transfer using omics data;
4.  ensure that resources are interoperable.

This consortium will make use of existing resources and developments, particularly FBA/RBA-type models (Goelzer et al., Gerlin et al., Colobié et al.), databases (MetExplore, ChloroKB) and quantitative data infrastructures (MetaboHUB, Atlas). The reflection and discussion will be generic: issues common to different modelling types (constrained and/or dynamic) will be addressed and will not be specific to a particular type of organism.

Over two years, the KOMOD consortium project plans to organize three workshops that will mobilize and adapt generic codes in pursuit of the above goals, as part of wider collaborations between software developers and biologists working on models of multicellular organisms (humans, animal organs or species, model plants or crops).

The short-term deliverables for the workshops will be the provision of operational guidelines for the acceleration of model construction and transfer, adding to the range of crops modelled (camelina and tomato, for example) and supporting projects that are either in progress or in the pipeline. They will also contribute to the standardization of data and models, reinforcing FAIR practices in production and on analytical platforms, and improving the interoperability of resources such as ChloroKB and MetExplore.

In the mid-term, the consortium will promote improved synergies in the use of data and modelling tools, more robust dataset frameworks to achieve successful transfer, and greater integration of the representation of biological processes to be shared among biologists, modelers and bioinformaticians.

Contacts:

Project participants

INRAE units