Figure 4

Workflow of the MDA2MSA algorithm. Step 1: Sequences with identical domain arrangements (a) are split according to domain boundaries (b). Then each part is separately aligned (c) and finally all parts are merged back together (d) into a single alignment. Step 2: The MDA (a) is used as a guide. The sequences are split into parts according to the MDA (b). Cuts are performed at the borders of aligned domains resulting in 5 parts. Each pair of sequence segments can now be aligned separately. In case unaligned domains occur in the MDA (part 3), the dynamic programming algorithm is changed such that it maintains the order of domains (c). The striped area represents the area that is not calculated because the MDA forbids the alignment of the two domains.