Objective: We tested an association rule mining algorithm in a biological database, in order to evaluate dependencies between lipids levels and a polymorphism (i.e., methylenetetrahydrofolate reductase gene mutation MTHFR 677C-T) in a sample of healthy adults of a Greek population.

Design: We studied genetic data from 322 men and 252 women who had no clinical evidence of cardiovascular or any other chronic disease. We measured total serum cholesterol, LDL cholesterol, oxidized LDL, as well as total plasma homocysteine concentrations and the distribution of the MTHFR genotype for all participants. Data analysis was mainly based on association rules to extract dependencies between the levels of biological factors and MTHFR genotype. These rules represent conditional probabilities of output as related to inputs.

Results: People with MTHFR TT genotype showed significantly elevated concentrations of total plasma homocysteine and oxidized LDL levels, when compared to the homozygous normal Results from association rule mining reveal the total unfavorable effect of the TT mutation in serum lipid profile.

Conclusion: The application of association rule mining may prove a useful tool in the analysis of large biological databases.