Pickaxe: a Python library for the prediction of novel metabolic reactions

Area of Science:

• Biochemistry
• Bioinformatics
• Computational Biology

Background:

• Biochemical reaction prediction tools are crucial for generating novel compounds and reactions.
• These networks have applications in designing biosynthetic pathways and annotating metabolomics data.
• Existing tools often lack flexibility, user-friendliness, and comprehensive reaction rules.

Purpose of the Study:

• To present Pickaxe, an open-source, flexible, and user-friendly software for generating novel reaction networks.
• To provide a robust method for tailoring reaction network generation to specific user applications.

Main Methods:

• Pickaxe iteratively applies reaction rules to a set of metabolites to generate novel reactions.
• Users can utilize pre-packaged rulesets (e.g., JN1224min from MetaCyc) or define custom rules.
• On-the-fly network pruning is enabled through filters based on chemical similarity, metabolomics, thermodynamics, and reaction feasibility.

Main Results:

• Pickaxe successfully generates novel metabolic reactions and compounds.
• Demonstrated applications include expanding biological databases, generating industrial chemicals from yeast metabolomes, and annotating E. coli metabolomics data.

Conclusions:

• Pickaxe offers a customizable and user-friendly solution for generating tailored metabolic reaction networks.
• The software is open-source, integrated into the MINE Database package, and supported by comprehensive documentation and examples.