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

Background: Accurate assignment of genes to pathways is essential in order to understand the functional role of genes and to map the existing pathways in a given genome. Existing algorithms predict pathways by extrapolating experimental data in one organism to other organisms for which this data is not available. However, current systems classify all genes that belong to a specific EC family to all the pathways that contain the corresponding enzymatic reaction, and thus introduce ambiguity. Results: Here we describe an algorithm for assignment of genes to cellular pathways that addresses this problem by selectively assigning specific genes to pathways. Our algorithm uses the set of experimentally elucidated metabolic pathways from MetaCyc, together with statistical models of enzyme families and expression data to assign genes to enzyme families and pathways by optimizing correlated co-expression, while minimizing conflicts due to shared assignments among pathways. Our algorithm also i...

• Type: Journal
• Year: 2005
• Where: BMCBI
• Authors: Liviu Popescu, Golan Yona
• Keywords: Algorithm, BMCBI 2005, Cellular Pathways, Metabolic Pathways