Applying genetic algorithms to the problem of variable selection in large datasets with interaction terms