The above optimisation problem makes sense only if the set Anc of anchor points is itself consistent. Since a user-defined set of anchor points cannot be expectd to be consistent, the first step in our anchoring procedure is to select a consistent subset of the anchor points proposed by the user. To this end, the program uses the same greedy approach that it applies in the optimisation procedure for multiple alignment. That is, each anchor point is associated with some user-defined score, and the program accepts input anchor points in order of decreasing scores – provided they are consistent with the previously accepted anchors.