In addition, we have a weak anchoring option, where consistent anchor points are only used to constraint the output alignment, but are not necessarily aligned to each other. More precisely, if a position x in sequence Si is anchored with a position y in sequence Sj through one of the anchor points, this means that y is the only position from Sj that can be aligned to x. Whether or not x and y will actually appear in the same column of the output alignment depends on the degree of local similarity among the sequences around positions x and y. If no statistically significant similarity can be detected, x and y may remain un-aligned. Moreover, anchoring x and y means that positions strictly to the left (or strictly to the right) of x in Si can be aligned only to positions strictly to the left (or strictly to the right) of y in Sj – and vice versa. Obviously, these relations are transitive, so if position x is anchored with position y1, y1 is to the left of another position y2 in the same sequence, and y2 in turn, is aligned to a position z, then positions to the left of x can be aligned only to positions to the left of z etc. The 'weak' option may be useful if anchor points are used to reduce the program running time.