Java scripts for implementation of the computational method for genetic discovery.

Identifying genes that are responsible for differences in metabolism could have important applications such as gene ranking for genome-wide association studies and drug target identification. We have developed a qualitative computational method that uses a curated metabolic network (YeastCyc) and LC/MS and GC/MS data of measured metabolite differences to identify genes that are likely to be causal for the metabolic differences.

The computational method first produces a table that predicts the changes in metabolite concentrations that would result from increasing or decreasing the activity of each metabolic gene. Second, based on this table and measured differences in metabolite concentrations between experimental and control samples, it computes a score that evaluates the potential for each gene to be a causative factor for those metabolic differences. The scores are based upon the quality of the match between the measured metabolic changes and the predicted effects in the table. The genes are then ranked by the scores, and the top-ranked genes are deemed to be mostly likely candidates for causing the differences in metabolites.

This collection has Java scripts demonstrating a implementation of this computational method. The readme.txt file shows how to run the scripts.