On the Assessment of Statistical Significance in Disease-Gene Discovery