Improving Metabolite Annotation through a New Generation of Parsing Tools for Genome-Scale Modeling

Authors: Dimitra Lappa, Benjamín J. Sánchez, Jens B. Nielsen

Facility for Computational Systems Biology

Department of Biology and Biological Engineering

Chalmers University of Technology

Improving metabolite annotation through a new generation of parsing tools for genome-scale modeling

The Human Metabolic Reaction (HMR) [1] is a genome-scale metabolic model (GEM) [2] that stands as a scaffold for integration of omics data and flux metabolic modeling of a generic human cell. HMR contains 3765 gene associations, over 8000 reactions and more than 3000 unique metabolites. A current challenge in GEM reconstruction is metabolite annotation, particularly for metabolites in cyclic reactions of lipid metabolism (e.g. fatty acid elongation), which usually lack unique identifiers so as to catalog into a database format. Here we describe a new strategy to correctly annotate these non-unique metabolites in the GEM modeling framework and specifically in the HMR model-derived database.

We present a new generation of parsing tools in the form of software plug-ins for a series of GEM file formats. Through parsing constraint-based model files we aim to unify nomenclature and introduce annotation for non-unique metabolites involved in HMR. The objective is to populate a schematic database to enable the querying of the model’s different parts. The parser’s input is a model in a popular format (xls, sbml, COBRA-sbml, RAVEN-sbml). The parser facilitates the conversion between formats and also breaks down the input files into model components. These components in turn populate a database containing unique metabolites, exchange reactions, reaction directionality, subcellular compartments and gene associations. The metabolites serve as the database’s primary index [3]. The parser will cope with nomenclature inconsistencies, converting annotation if necessary. Importantly, it will detect fatty acids cyclic reactions and attempt to label them accordingly. Complementary checks will subsequently be carried to avoid structural inconsistencies and ensure the model’s integrity. This approach is expected to help towards model standardization, while addressing the challenge of cyclic reactions in modeling the cellular metabolism and promoting the comparison and evaluation among different models.

Keywords:          genome-scale metabolic models, metabolic annotation, non-unique metabolic identifiers, human reaction database, parser

Topic:                 New metabolic models and simulation algorithms

References:

[1] Mardinoglu A., Agren R., Asplund A., Uhlen M., Nielsen J.B.,“ Genome-scale metabolic modeling of hepatocytes reveals serine deficiency in patients with non-alcoholic fatty liver disease”, Nature Communications 5, 3083, 2014

[2] Lewis, N.E. et al., “Constraining the metabolic genotype–phenotype relationship using a phylogeny of in silico methods”, Nature Reviews 10, 291-305, 2012

[3] Pornputtapong N., Nookaew I., Nielsen J.B., “Human Metabolic Atlas: an online resource for human metabolism”, Database: The Journal of Biological Databases and Curation, In Press