Abstract
AbstractMotivationThe metabolic network of an organism can be analyzed as a constraint-based model (CBM). CBM analyses can be biased, optimizing an objective such as growth rate, or unbiased, attempting to describe the feasible flux space through pathway analysis or random flux sampling. Biased methods scale to genome-scale models (GEMs) and beyond, but unbiased methods suffer from combinatorial explosion. One promising approach to improve the scalability of unbiased analyses is to focus on cells’ metabolite exchanges with each other and the environment rather than the internal details of their metabolic networks. However, methods for unbiased analysis of metabolite exchanges have not been evaluated or compared.ResultsUsing CBMs ranging in size from core models to GEMs, we applied pathway enumeration and random flux sampling to metabolite exchanges in microbial species and a microbial community. We focused on pathway definitions that allow direct targeting of metabolite exchanges: elementary conversion modes (ECMs), elementary flux patterns (EFPs), and minimal pathways (MPs). Enumerating growth-supporting ECMs, EFPs, and MPs for the same models, we found that the MPs were always a subset of the ECMs and EFPs, specifically the subset of support-minimal pathways. The EFPs were always a subset of the ECMs, which in turn were always a subset of flux patterns from elementary flux modes (EFMs). We argue that this hierarchical relationship between pathway definitions holds more generally and that it helps explain the relative scalability of ECM, EFP, and MP enumeration. Metabolite exchange frequencies in enumerated pathways were similar across methods with a few specific exchanges explaining large differences in pathway counts. Thus, biological interpretation was largely robust to the choice of pathway definition. Random flux sampling scaled to GEMs and complemented pathway enumeration.AvailabilityData and code are available athttps://gitlab.com/YlvaKaW/exchange-enumeration.
Publisher
Cold Spring Harbor Laboratory