The accuracy of progressive methods is strongly influenced by the quality of initial alignment and guide tree. Refinement on the initial alignment and guide tree can significantly reduce the estimation error of progressive methods. This idea motivates the iterative methods which rebuild the guide tree and MSA iteratively.
MUSCLE (multiple sequence alignment by log-expectation) is a popular iterative method for multiple sequence alignment. It updates the guide tree with a more accurate distance measure based on the new MSA between iteration stages. Then the updated guide tree is used to build a more accurate MSA.
A list of programs implementing iterative methods