Most of the existing algorithms for mining frequent patterns could produce lots of projected databases and short candidate patterns which could increase the time and memory cost of mining. In order to overcome such shortcoming, we propose two fast and efficient algorithms named SBPM and MSPM for mining frequent patterns in single and multiple biological respectively. We first present the concept of primary pattern, and then use prefix tree for mining frequent primary patterns. A pattern growth approach is also presented to mine all the frequent patterns without producing large amount of irrelevant patterns. Our experimental results show that our algorithms not only improve the performance but also achieve effective mining results.