Toward the Reconciliation of Inconsistent Molecular Structures from Biochemical Databases

Area of Science:

• Biochemistry and computational biology
• Metabolomics and systems biology
• Drug discovery and cheminformatics

Background:

• Molecular structure data is crucial for metabolomics, systems biology, and drug discovery.
• Existing biochemical databases often lack completeness and present inconsistent structural information for compounds.
• Integrating data from multiple sources is necessary but challenging due to data discrepancies.

Purpose of the Study:

• To present StructRecon, a novel computational tool designed to resolve unique molecular structures from database identifiers.
• To address the challenge of inconsistent molecular structure data across different biochemical databases.
• To provide a standardized and more complete representation of molecular structures for downstream applications.

Main Methods:

• StructRecon constructs identifier graphs by traversing cross-links between entries in supported databases (BiGG, ChEBI, ECMDB, MetaNetX, PubChem).
• An extensible model for molecular structure, supporting multiple levels of detail, is developed to reconcile discrepancies.
• Standardization is applied iteratively, and a random walk-based algorithm selects the most likely structure when multiple candidates arise.

Main Results:

• StructRecon successfully integrates information from multiple biochemical databases to resolve unique molecular structures.
• The tool provides a more comprehensive view of compound information by reconciling inconsistencies.
• The developed extensible model and random walk algorithm effectively handle structural discrepancies and select the most probable structure.

Conclusions:

• StructRecon offers a robust solution for obtaining consistent and unique molecular structures from diverse database identifiers.
• The tool enhances data integration capabilities in fields reliant on accurate molecular structure information.
• StructRecon facilitates more reliable analyses in metabolomics, systems biology, and drug discovery by providing standardized structural data.