A new network representation of the metabolism to detect chemical transformation modules

Abstract

Background: Metabolism is generally modeled by directed networks where nodes represent reactions and/or metabolites. In order to explore metabolic pathway conservation and divergence among organisms, previous studies were based on graph alignment to find similar pathways. Few years ago, the concept of chemical transformation modules, also called reaction modules, was introduced and correspond to sequences of chemical transformations which are conserved in metabolism. We propose here a novel graph representation of the metabolic network where reactions sharing a same chemical transformation type are grouped in Reaction Molecular Signatures (RMS).

Results: RMS were automatically computed for all reactions and encode changes in atoms and bonds. A reaction network containing all available metabolic knowledge was then reduced by an aggregation of reaction nodes and edges to obtain a RMS network. Paths in this network were explored and a substantial number of conserved chemical transformation modules was detected. Furthermore, this graph-based formalism allows us to define several path scores reflecting different biological conservation meanings. These scores are significantly higher for paths corresponding to known metabolic pathways and were used conjointly to build association rules that should predict metabolic pathway types like biosynthesis or degradation.

Conclusions: This representation of metabolism in a RMS network offers new insights to capture relevant metabolic contexts. Furthermore, along with genomic context methods, it should improve the detection of gene clusters corresponding to new metabolic pathways.

Fig. 1

Reaction network to Reaction Molecular Signature network. This figure presents a toy example of the reduction of a reaction network in a RMS network. Reactions sharing a same reaction signature (same node color in the figure) are grouped in a single RMS node. Directed edges of the reaction network are also merged in the RMS network. Red edges illustrate the computation of Markov transition probabilities P r(R M S ₂∣R M S ₁), P r(R M S ₃∣R M S ₁) and P r(R M S ₅∣R M S ₁). They correspond to the proportion of reaction edges, among the five outgoing edges of RMS1 reactions (blue nodes), connecting RMS1 to RMS2, RMS3 and RMS5

Fig. 2

Example of reactions having a same RMS signature but classified in different EC classes. a D-glutamate cyclase reaction annotated with the EC 4.2.1.48. b L-lysine lactamase reaction annotated with EC 3.5.2.11. This both reactions make the same the chemical transformation represented by RMS-H1.1372, which encodes, in SMILES-like strings, the difference between the products and the substrates of atomic signatures of height 1

Fig. 3

Conservation of β-oxidation module for non-fatty acid compounds. In addition to fatty acids, the β-oxidation module was found conserved for the transformation of 8 compounds represented in the figure. For the first step, we found 4 reaction variants encoded in different RMS of height 1: three RMS correspond to a dehydrogenation between the alpha and beta carbons but with different acceptors, another corresponds to a coenzyme A ligation. A color code indicates the corresponding substrates. Only molecules marked with an asterisk were also detected by Muto et al. (KEGG Reaction Module RM018)

Fig. 4

A conserved module for the biosynthesis of aldoximes from amino acids. a This module is made of three chemical transformations encoded by RMS-H2 signatures. It corresponds to the oxidative decarboxylation of an anmino acid to its aldoxime. b The module is conserved in different MetaCyc pathways for five distinct proteinogenic amino acids. Produced aldoximes are precursors of nitrogen-containing secondary metabolites in plants, like cyanogenic glycosides for seed germination and defense, or auxin phytohormones

Fig. 5

Boxplots of conservation scores for enumerated and known metabolic paths. For paths of length 2 (two edges and three nodes) in the RMS-H2 network, distributions of the three conservation scores (i.e. scoreRea, scoreProt and scorePageRank) are presented in all possible paths from the RMS network (identified as “All paths” in the figure) versus paths solely included in known metabolic pathways (“Known metabolic pathways”). The latter present significant higher scores (p-value <2e ⁻¹⁶ using Tukey’s HSD tests)