Reconstruction refinement pipeline DEMETER

[Almut Heinken]

Dear COBRA Toolbox users,

I would like to introduce a new COBRA Toolbox extension that has recently been integrated: the reconstruction refinement tool DEMETER (Data-drivEn METabolic nEtwork Refinement). DEMETER enables the simultaneous refinement of many reconstructions at once based on comparative genomics and experimental data. DEMETER was previously used to reconstruct 818 human gut microbial reconstructions, AGORA (https://www.nature.com/articles/nbt.3703), and its expansion, AGORA2, accounting for 7,206 strains (https://www.biorxiv.org/content/10.1101/2020.11.09.375451v1). The starting point is a draft reconstruction generated through the KBase framework (kbase.us/).

A key feature of DEMETER is that it performs expansion by species-specific pathways based on experimental data that was manually collected for over 1,000 species. The experimental data can be propagated to new strains of existing species. Otherwise, newly collected data can easily be filled in and read by the pipeline. Comparative genomics data can also be read in and pathways and genome annotations of the reconstructions will be refined accordingly.

Other functions carried out by DEMETER include translation from KBase to VMH (https://www.vmh.life/) identifiers,  ensuring that the resulting refined reconstruction is thermodynamically feasible and can grow anaerobically, curation of the biomass objective function based on gram status, and construction of a periplasmatic compartment where applicable.

DEMETER can be rapidly carried out for many reconstructions simultaneously. I have created a tutorial that explains the input parameters at https://github.com/opencobra/COBRA.tutorials/tree/master/reconstruction/demeter. A test and debugging suite will also be carried out simultaneously for all reconstructions.

Another function in DEMETER is that it enables the computation of model properties, such as distribution of reactions, and uptake and secretion potential. The results can then be clustered to visualize the similarities between organisms. This can also be done for the already available AGORA reconstructions. The tutorial also explains this step.

I would be happy to answer any questions about how to run DEMETER.

Kind regards,

Almut