Mining Gene Sets for Measuring Similarities