GPRuler: Metabolic gene-protein-reaction rules automatic reconstruction

Area of Science:

• Computational biology
• Systems biology
• Bioinformatics

Background:

• Metabolic network models are vital in healthcare and industry.
• Automated reconstruction tools exist, but gene-protein-reaction (GPR) rule generation is manual and time-consuming.
• GPR rules are essential for gene deletion simulations and integrating gene expression data.

Purpose of the Study:

• To develop an open-source Python framework, GPRuler, for fully automating GPR rule reconstruction.
• To enable rapid and resource-efficient GPR reconstruction for any organism.

Main Methods:

• GPRuler mines text and data from nine biological databases.
• It reconstructs GPRs from an organism's name or an existing metabolic model.
• Performance was evaluated on small-scale and genome-scale metabolic models (Homo sapiens, Saccharomyces cerevisiae).

Main Results:

• GPRuler accurately reproduced original GPR rules in benchmark models.
• The tool demonstrated high accuracy, often exceeding existing models.
• Manual investigation revealed GPRuler's potential for improved accuracy.

Conclusions:

• GPRuler significantly automates GPR rule reconstruction, saving time and resources.
• It facilitates the study of context-specific metabolic networks.
• The tool supports metabolic network analysis for industrially or biomedically relevant organisms.