Pan‐Genome‐Scale Network Reconstruction: Harnessing Phylogenomics Increases the Quantity and Quality of Metabolic Models

Abstract

A genome-scale network reconstruction (GENRE) is a knowledgebase for an organism and has various applications. Available genome sequences have risen in recent years, but the number of curated GENREs has not kept pace. Existing yeast GENREs contain significant commission and omission errors. Current practices limit the quantity and quality of GENREs. An open and transparent phylogenomic-driven framework is outlined to address these issues. The method is demonstrated with 33 yeasts and fungi in Dikarya. A pan-fungal metabolic network called FYRMENT (Fungal and Yeast Metabolic Network) (https://github.com/LMSE/FYRMENT) is created, and annotated with ortholog groups from AYbRAH (https://github.com/LMSE/AYbRAH). Metabolic models for lower-level taxons are compiled. The fungal pan-GENRE contains 1553 orthologs, 2759 reactions, 2251 metabolites. The GENREs have higher genomic and metabolic coverage than existing yeast and fungal GENREs created with other methods. Metabolic simulations show the maximum amino acid yields from glucose differs between yeast lineages, indicating metabolic networks have evolved. Curating genomes and reactions at higher taxonomic-levels increases the quantity and quality of GENREs than conventional approaches. This approach can scale to other branches in the tree of life.

Yeast metabolic innovations emerged via expanded metabolic network and gene positive selection

“…We firstly reconstructed a yeast pan‐GEM using a new pipeline (Materials and Methods) developed based on the template model strategy (Machado et al , 2018; Correia & Mahadevan, 2020) (Fig 1, Materials and Methods). The pan‐GEM comprises of metabolic reactions and enzymes from all 343 fungal species, containing a total of 3,135 metabolites, 4,599 reactions and 3,751 ortholog groups, which therefore represents a significant expansion of coverage in metabolism compared with a prior fungal pan‐GEM (Correia & Mahadevan, 2020) (Appendix Fig S1J). The ssGEMs for all 343 fungal species were initially generated based on the existence of enzyme orthologs as annotated in the pan‐GEM, then followed by gap‐filling (Appendix Fig S1K, Materials and Methods).…”

“…Three criteria were applied for evaluating reactions to be included in the pan-GEM: (i) unbalanced, reactions with generic reactants (e.g. "sugar") and reactions containing "n" in the stoichiometry were discarded; (ii) new reactions that occurred in a previously reported pan-fungal GEM (Correia & Mahadevan, 2020) were included; and (iii) reactions with more than 2 dead-end metabolites were initially filtered out. For the new reactions, HMM-based gene associations from KEGG, EggNOG and RAVEN-KEGG were prioritized in comparison with the approach of homology search used by RAVEN-MetaCyc.…”

Genome-scale modeling of yeast metabolism: retrospectives and perspectives

“… 2019 ). In addition, species-specific GEMs for 33 yeasts and fungi were also reconstructed, which predicted different amino acid yields from glucose and therefore reflected evolution of metabolic networks (Correia and Mahadevan 2020 ). Moreover, the large-scale reconstruction of species-specific GEMs for 343 fungal species enabled comprehensive analyses of evolutionary diversification of substrate utilization (Lu et al .…”

AuCoMe: inferring and comparing metabolisms across heterogeneous sets of annotated genomes

“…This is relevant for organisms for which multiple GSMNs exist, in order to establish a reference one. This strategy was successfully applied to Salmonella Typhimurium LT2 (Thiele et al , 2011) as well as Saccharomyces cerevisiae (Herrgård et al , 2008), and was later extended to multiple organisms to create a panmetabolism of 33 fungi from the Dikarya subkingdom (Correia and Mahadevan, 2020). Although platforms now facilitate such community efforts (Cottret et al , 2018), these methods are costly in terms of time and manpower involved.…”