Based on the mixed idea of Progressive alignment and Divide-and-conquer alignment, two different multithreaded multiple sequence alignment programs, depending on how the guide tree(s) would be applied, are implemented for checking the improvements of alignment speed and sensitivity. The single-tree alignment builds a uniformed guide tree for the full-length sequences at the beginning, which will be used by all the sub-alignments as guide tree. In multiple-tree alignment, sequences will be firstly cut into pieces and these sub-sequences will build their own guide trees to guide their individual alignments. Multiple-tree alignment seems having a better speedup performance than the single tree alignment, but neither of them, at this stage, shows ideal sensitivity results as the number of threads increases. Therefore, some heuristic methods for fixing the cut points were suggested for future improvement, such as overlapping alignment and sliding window alignment.