The neighbor-joining (NJ) tree method, introduced in 1987 by Naruya Saitou and Masatoshi Nei, is a distance-based method for constructing evolutionary trees. The NJ algorithm removes molecular clock assumptions, calculating distances from an internal node to each of the respective leaves and results in an un-rooted tree. The algorithm attempts to minimize the sum of all branch-lengths on a constructed tree, beginning by forming a star-like tree, with each leaf corresponding to a particular sequence. As the program progresses, it iteratively picks two nodes adjacent to the root and joins them by inserting a new node between the root and the two selected nodes. For the original paper concerning Neighbor-Joining Trees click here.
The tree building program returns the evolutionary tree, along with clustering and distance information in PHYLIP format. For additional information concerning PHYLIP and further processing of the data produced continue here.