I have a set of protein sequences which I have aligned by using different Multiple Sequence Alignment methods such as TM-Coffee, MUSCLE, multialign in MATLAB, etc. Since this set of sequences is very divergent the alignment I am getting from different tools is very different so I am searching for a way to know which of the alignments is better. There are some papers which have looked at the accuracy of the alignment methods individually but I want to have a score for each alignment and compare that to arrive at a decision. Would the sum of pairs scoring work? Any papers that have used it?