Automation of gene assignments to metabolic pathways using high-throughput expression data

Area of Science:

• Genomics
• Systems Biology
• Bioinformatics

Background:

• Accurate gene-to-pathway assignment is crucial for understanding gene function and mapping genomes.
• Current methods extrapolate data across organisms and often assign genes ambiguously to multiple pathways.
• Existing systems lack specificity, assigning all genes in an EC family to all associated pathways.

Purpose of the Study:

• To develop a novel algorithm for the selective and accurate assignment of genes to cellular pathways.
• To address and reduce the ambiguity inherent in current gene pathway assignment methods.
• To improve the functional annotation of genomes by refining gene-pathway relationships.

Main Methods:

• Developed a new algorithm integrating experimental pathway data (MetaCyc) with statistical models of enzyme families and gene expression data.
• Optimized gene-to-pathway assignments by maximizing correlated co-expression and minimizing assignment conflicts.
• Incorporated identification of alternative genes and handling of multi-domain proteins.

Main Results:

• The algorithm selectively assigns specific genes to pathways, reducing ambiguity.
• Applied to the Yeast genome, assignments showed consistency with experimentally verified data.
• The model reflects characteristic properties of cellular pathways and improves accuracy.

Conclusions:

• An algorithm for automatic assignment of genes to metabolic pathways has been developed.
• The algorithm leverages gene expression data to decrease ambiguity associated with EC number-based assignments.
• This approach enhances the precision of functional genomics and pathway analysis.